👉 Fill out the listener survey - https://services.multiplai.ai/lai-survey
Is the internet as we know it about to vanish?
Recent data shows a dramatic shift—9 out of the top 10 websites are losing traffic while AI tools like ChatGPT are skyrocketing. The era of web browsing is transforming, and if your business relies on old-school web traffic, you might be sitting on a ticking time bomb.
In this powerful news breakdown, we explore the most shocking AI stories of the week, including how AI is not only disrupting the internet but also reshaping higher education and even influencing courtroom decisions with digital avatars.
The big question: Are you ready for the AI-first world?
In this session, you'll discover:
- The End of Traditional Web Browsing: Why traffic to major websites is plummeting and what that means for your business.
- The AI Avatars in Courtrooms: How digital representations are now being used to influence legal decisions—and the ethical chaos that follows.
- Higher Education’s AI Crisis: Students are using AI to complete up to 90% of their assignments. What does that mean for the workforce of tomorrow?
- The Shift in Global Wealth: Why AI could trigger a $100 trillion power shift in the global economy.
- Saudi Arabia's $600B AI Bet: How new investments are setting the stage for US jobs and AI infrastructure—and the risks of foreign control.
About Leveraging AI
- The Ultimate AI Course for Business People: https://multiplai.ai/ai-course/
- YouTube Full Episodes: https://www.youtube.com/@Multiplai_AI/
- Connect with Isar Meitis: https://www.linkedin.com/in/isarmeitis/
- Join our Live Sessions, AI Hangouts and newsletter: https://services.multiplai.ai/events
If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Isar Meitis (02:09):
Hello and welcome to a Weekend News episode of the
Leveraging AI Podcast, thepodcast that shares practical,
ethical ways to improveefficiency, grow your business,
and advance your career.
This is Isar Meitis, your host,and we are going to cover some
very profound and interestingtopics that happened this past
week.
The first one, we're going totalk about the first signs of
the end of the internet as weknow it.
(02:31):
Yes, I know that sounds big, butthat's the reality.
The second topic that we'regoing to talk about is how
impactful is AI in highereducation right now, and how it
is going to change highereducation and the workforce in
the next few years, which isgoing to impact all of us.
And the third main topic isgoing to be Trump's visit in
(02:52):
Saudi and how that is connectedto AI and AI infrastructure.
But before we begin, there's twothings I want to add.
First of all, next week I'mgoing to do a special deep dive
in the first section of theepisode, instead of the regular
weekly things I'm going to diveinto the AI Company of the
Future.
It's a podcast that was releasedby dsh who has an amazing
podcast that you should followif you're not following already.
(03:13):
he is a brilliant mind and he'sa great interviewer and he has
access to anybody he wants inthe AI world.
But he was doing actually a soloepisode talking about the AI
company of the future.
You can go and listen yourself,but I'm gonna cover it, in depth
next week.
I wanted to do it this week, butthere was too many important
things to talk about, which Ihope I'll be able to avoid, next
week and jump into that.
And the last thing I wanna saybefore we start is I wanna say
(03:35):
thank you.
so first of all, thank you toall of you for listening to this
regularly, but it's amazing tome how close this community is
becoming of people followingthis podcast and engaging with
me across multiple channels.
We have now several differentweeks, more than 20 people every
week on the AI Friday Hangouts,which you are more than welcome
to join, we'll do this at everyFriday at 1:00 PM Eastern.
(03:55):
If you wanna join, there's alink to do that in the show
notes.
But I want to give a specialthing to Charlie Hoing, Leon
Crisp and Dennis Kovich.
All three of them heard anepisode I released a couple of
weeks ago saying that I like toget an access to Manus and I
don't have access yet.
And all three of them reachedout and offered an invitation
from them to give me access tothat, which I now have and can
(04:17):
experiment and try differentthings, which I will report in
an episode in the next couple ofweeks.
So thank you all of you forlistening.
Thanks specifically to Charlie,Leon and Dennis for caring and
for offering me Manusinvitations.
And now let's get started withthe episode.
(04:37):
So I mentioned the first topicis the end of the internet as we
know it.
I talked about this in manyprevious episodes on how AI is
changing the way we engage withinternet, and how agents and
search are gonna be impacted bythat, and how my personal
behavior has changeddramatically, where probably 85%
of what I'm looking for is notcoming from going to Google, but
(04:58):
going through different AItools.
But this week there was a crazeon Reddit on the data about the
traffic to the top 10 leadingwebsites in the world.
This is the first time in maybethe history of the internet, but
definitely in recent years thatthe traffic to nine out of the
top 10 websites has decreasedmonth over month, and there's
(05:20):
only one website that saw anincrease in the top 10, and that
is ChatGPT.
That, by the way, was number 15on the list just a year ago, and
now it is in number five andgrowing.
What does that tell us?
Well, first of all, I need tosay that this is not one data
point, but the one data point isvery significant.
The regular traffic to websitesis declining.
(05:41):
Apple also reported the overalltraffic to websites and browsers
is declining with more and moreusers globally are now using
ChatGPT Gemini, and other appsto research things that they're
trying to find.
This, as I mentioned again inprevious episodes as well, is
going to have profoundimplications on every online
(06:03):
presence of any company.
If the future of your businessdepends on website traffic as it
looks today, even if it's 30% ofthe revenue that you're making,
40% of the revenue that you'remaking, if you're not going to
make dramatic changes in thenext few years, that is going to
go away because the internet isgoing to be browsed through
agents.
You will talk to whateverapplications, whether it's
(06:25):
ChatGPT, Gemini, Claude Grok, oranything else that will come out
in the next few years, and thesewill do the research for you and
we'll get you back information.
And the companies that willunderstand how to build the
backend of their website in away that will be agent friendly
will win.
And anybody who is late to thatgame will lose potentially their
entire business.
(06:46):
I'm having conversations withvery senior people in really
large companies, in the billionsof dollars who are not aware of
what I just said, and they'renot planning yet.
That tells me two things that,one, the understanding that
people at the leading edge of AIhas right now that is very
obvious to them is not even onthe radar of most companies in
(07:06):
the world.
And the other thing is thatthere's a huge risk and
opportunity of making the rightmoves to understand how to
engage with agents.
To be fair, it's not awell-defined path,.Meaning it's
not very clear how that's gonnaevolve and how it's gonna look
like in two to three years.
But what is very obvious to meis that it's not going to be
what we see and know right now,or what we see and know and
(07:27):
everything we've learned fromrunning web-based businesses in
the past 20 years.
Now, I said that was just onedata point, but the second data
point comes from an interview byEd DQ from Apple, who revealed
potential plans to shift Safarisearch from Apple to AI tools
like Open ai, perplexity,Anthropic, and so on.
That one sentence sent Alphabetsshares down 7%.
(07:52):
That's$150 billion in marketvalue that were eliminated
because of that.
Q'S specific quote was, AIservices will eventually replace
standard search engines such asGoogle.
Now he's Apple's vice presidentof services..
He doesn't just say things.
I think he was very obvious tohim what's gonna be the
implications of what he'ssaying.
And the other piece ofinformation is that Safari
(08:13):
searches fell for the first timein April of 2025.
So again, another data pointthat shows you where is the
direction going, A, from thevolume of search to traditional
search, and B, from the plans ofone of the largest companies in
the world when it comes togiving people access to data,
meaning Apple in all of itsdevices.
(08:34):
combine that with the US JusticeDepartment lawsuit that is
trying to break up Google andpotentially disallow Apple to
continue using Google as itsprimary engine, as well as
potentially forcing Google tosell Chrome.
And you see that things aregoing to change potentially
faster than we initiallythought.
Now if that's not a big enoughblow to your brain to think
(08:54):
about.
I wanna add a few more points,and I mentioned those in
previous episodes, but I wannatouch on them again because I
think they're critical and maybeyou didn't listen to all the
episodes.
the first one is, what happensto the financial engine of the
internet when, and if thishappens.
Meaning, the first one is whatis going to happen to the
internet from a financialperspective when this transition
happens?
(09:14):
So what I mean by that is, thinkabout it.
The internet is the mostincredible service, potentially
the human kind ever had.
That is free.
None of us is paying for theinternet, mean we are paying the
ISP to get internet access atour house.
But we're not paying for theinternet itself.
And the reason for that isbecause ads are paying for the
internet.
But if I'm using ChatGPT orGoogle or Gemini or Grok or
(09:36):
Manus or whatever the case maybe to access the internet, it's
not seeing ads.
I don't see the ads.
I just see the answers that Iwas looking for.
It's gonna book the tickets forme that I needed.
It's gonna order the food that Iwant and so on.
And that means there are no ads.
No ads, no money, no money whopays for the internet.
And I don't think anybody hasanswer to that yet.
(09:57):
The second question is.
Why would people create newcontent if there is no
incentive?
Meaning the incentive to createnew content that gave us
information was I create qualitycontent that's gonna get found
by Google, that's gonna sendpeople to my website, that will
connect with me, my brand, mycompany, and will buy our
products and services.
That model is going to change aswell, which means the incentive
(10:20):
to create new valuable contentmay or may not be there,
depending how this evolves.
So the question becomes who orwhat is going to create new
content in the future?
And that's another question Idon't think anybody has answer
to.
A lot of it is gonna be AIcreated, but even then you're
asking, okay, what kind ofcontent will it create for what
purposes, based on whatinformation and so on.
(10:40):
And the third question, which ismaybe the most important
question is how do you in yourbusiness, in your strategy, are
preparing for this future whenthe facts of how it's going to
evolve are unclear?
And I think the answer for thatis you gotta stay.
And I think the answer for thatis you gotta start researching
how it's gonna impact yourindustry, your business, your
(11:01):
universe, and in parallel, be ontop of what is changing in the
agent world to slowly try tofigure out and try to make the
changes faster than yourcompetition.
So you can capture market shareas this shift happens.
It's not a shift that's gonnahappen overnight, it's gonna be
a process.
And this process may take 18months to five years, but during
(11:22):
that timeframe, there's gonna bemore and more traffic that is
gonna be agent based versushuman based.
the second topic that we'regoing to dive into is the impact
of AI on higher education.
A very long and detailed articlein the New York Times has shared
findings on research thatthey've done on the impact of AI
in higher education.
And what is happening in highereducation right now is the
(11:44):
younger generation that is inuniversities right now in that
is in colleges and universitiesin the US right now is using
ChatGPT and other AI tools inmore and more ways with more and
more students claiming thatthey're generating 70, 80, 90%
of their work with ai, with themjust adding 10 to 20% of the
(12:05):
effort in order to get, to gothrough, in order to go through
their TA in order to completetheir, in order to complete the
university.
In order to complete the taskthat their professors are giving
them.
Now, this has many differentimpacts, both on students and
professors.
First of all, the students'integrity is at question, but it
(12:28):
just depends on what the rulesare in each university or in
each professor and each roomwith one of them defines it
differently.
But it's very, very clear thatmore and more students are using
AI tools in order to do the workfor them and not disclosing it
as they're doing that.
And yes, there are attempts bydifferent universities to have a
more holistic approach to thisphenomenon, but it's definitely
(12:49):
not the norm and it's definitelynot a standard with every
university and sometimes everyprofessor doing their own thing.
Now to tell you how crazy thesituation is, one of the
examples is a philosophyprofessor at the University of
Arkansas.
,Little Rock caught studentsusing AI to write personal
introductions for an ethics andtechnology class.
So I will let that sink for aminute.
(13:10):
It's a personal introduction.
You're telling about yourself toother student in an ethics
class, and there were studentsusing AI to write that piece of
information.
Now, that's funny, but that'salso really, really scary.
It is scary because what kind ofcritical thinking capabilities
are these students developing?
If every task that they need tocomplete is actually done by ai.
(13:31):
Now it's scary year.
If you think about that, theseare the people that are going to
come into the workforce twoyears from now, because in the
past two years, every task thatthey had, they've done with ai.
How are they going to respondwhen they get into a company in
which they cannot use AI foranything?
Well, they may not know how todo projects without AI because
that's everything they've donefor their four years in college
(13:52):
and university that was supposedto prepare them for working in
the companies.
Will they come and work incompanies that do not have very
open AI policies?
What kind of policies you needto have in place as a company
once you start receiving thesepeople into the workforce, that
their solution for everything isChatGPT or any other AI tool.
Now, professors are reallyfrustrated with the whole
(14:14):
situation again because theydon't have a good answer to how
to address the currentsituation.
In many cases, the universitiesare asking them to grade work of
students that was not done bythe students themselves, but was
done by ai.
It's very obvious to them thatthat is the situation, and yet
they still have to grade itbecause the university currently
does not have a better solution.
And yes, there are some AIcheckers, but it's almost as
(14:36):
accurate as flipping a coin,meaning whether you find that
something was done by AI or not,with these checkers, you still
don't know for sure.
And then what are you going todo?
You're gonna throw somebody offtheir university degree when
it's a 50% chance, 70% chance,90% chance.
Where do you draw the line inthe sand?
If you're saying, I'm gonna messwith somebody's future when I'm
not a hundred percent sure thatthey're actually cheating.
The flip side of that is.
(14:57):
Is it cheating?
Are we really preparing them forthe future if we're not allowing
them to use ai?
So I think this is on one hand,a really scary article that
raises a lot of concerns andquestions.
But on the other hand, it isvery obvious that it's the
biggest opportunity theeducation system as a whole ever
had.
because if you think about howwe are teaching students right
(15:17):
now, whether in elementaryschool, middle school, high
school, or higher education.
It is very, very similar to howwe've been teaching for hundreds
of years.
I live in Florida, not too farfrom me.
There's a beautiful town calledSt.
Augustine.
St.
Augustine has the first woodenschool in the United States.
It was built sometime between1702 and 1716, so that's a very
(15:38):
long time ago.
Over 300 years.
And you know what it has aclassroom with desks and chairs
and a blackboard and a place fora teacher to stand in front of
the students.
That looks and feels very muchlike what we're doing today.
So we have an opportunity todeliver personalized, high
quality training and educationto every kid with the teacher
(15:59):
and or professor being more of amentor that can provide the
human side, the mentorship, thehandholding, the moral and
ethical support for studentsversus the actual teaching
itself, which I believe AI canpersonalize and do better than
what a teacher or a professorcan do in front of a class of 30
to 400 students, depending onthe size of the class in the
specific facility.
(16:19):
How do we then balance betweencritical thinking, AI tools and
other tools that will come up inthe future?
I don't know, but I think itrequires and hopefully will
force a complete revolution ofour education system.
And it doesn't end withstudents.
The problem with professors isas profound, meaning professors
are stuck in this weirdsituation when they themselves
(16:40):
are using more and more AIbecause it's helps them create
better lessons, faster andcheaper, faster and easier.
And on the other hand, they'renot allowing their students to
use it.
But an interesting example camethis week from Ella Stapleton, a
2025 nor Northeastern Universitygraduate, who's demanding the
university to pay her an$8,000tuition refund after she
(17:00):
discovered that her businessprofessor Rick, Arrow would re
arrow would use ChatGPT togenerate lectures and lecture
notes without disclosing it.
So on the student side, there'sthe same level of frustration,
right?
Paying all this amount of moneyin many cases with huge student
loans.
And now only to find out that myprofessor, instead of providing
me the information that helearned through the years, he's
(17:21):
actually using ChatGPT toprovide me information that I
could have gotten for free.
And this obviously doesn't endin the universities because it
immediately reflects oninternships and entry level
jobs.
In a recent report by the WorldEconomic Forum, it was found
that AI automation is takingover entry level jobs and it
reducing traditional internshipson opportunities in a very
(17:43):
dramatic way.
Several different professorsfrom known universities around
the world in that research areemphasizing the role of
preparing students for careersusing internships, but they're
claiming that it's becomingharder and harder for them to
find internships for students inspecific areas.
This just amplifies the impactof the education problem itself.
(18:04):
Meaning, on one hand, thestudents are going to do more
and more of their work with AIwithout actually developing some
of the skills they need for thejob force.
On the other hand, they can'teven learn that on the job as
part of an internship becauseinternships are becoming more
and more rare because companiesare saying, I don't need an
intern.
I can actually do this with ai.
And that obviously stops theentry for the waterfall effect,
(18:24):
that then needs to become moresenior employees in multiple
companies.
I asked that question multipletimes.
If you don't need juniordevelopers, how do you get
senior developers?
How do you get team leads?
How do you get CTOs after that?
And the same question happens inlaw firms, et cetera, et cetera.
These are a lot of questionsthat I don't think anybody has
answers to at this point.
Now, the flip side that I do seeis I see a big opportunities for
(18:45):
companies to be able to shift toAI usage because they will hire
people straight out ofuniversity that have significant
access and experience to usingAI tools across multiple aspects
of things that they had to doduring the university.
So while you have a lot ofemployees who are used to doing
things the way they're used todoing it, you may have a breath
of fresh air when it comes tohiring people out of university
(19:08):
who has been using AI everysingle day for four years, which
is probably not the case forpeople in your business.
So there is a light at the endof the tunnel.
I just don't know how long thetunnel is, and I am not sure how
long it will take us to alladjust to this new situation
because it's shifting very, veryfast.
And the bigger systems such asthe education system, the
medical system, the legal systemand so on, are very slow to act
(19:32):
and it will be a turbulent timethrough this transition period.
Now, on the same topic of AIImpact on the Future economy, a
very interesting article on FastCompany addressed actually two
aspects of a) what they call the100 trillion power shift.
And they're saying two verypowerful things are going to
happen in the global economy inthe next 20 years.
(19:52):
One is the rise of AI that, as Imentioned, they're claiming can
drive a hundred trillion dollarsrelocation in global wealth by
2043.
Basically just short of 20 yearsdoubling the global economy,
fueled by automation andinnovation, what they're
arguing, which we've heard timeand time again that AI is not
just going to take jobs, butit's going to actually help
(20:14):
offsetting the shrinkingworkforces in many countries
around the world, especially inaging nations like Japan.
So I highly recommend, if youhaven't read the book, the End
of the World is just thebeginning that you go and read
that book, or at least listen toit if you're more of an
audiobook kind of person.
It's a fascinating book thattalks in addition to a lot of
other things about the shrinkingworkforce in several different
(20:35):
places around the world and theimplications it will have on
these particular economies.
So being able to offset thatwith AI and then robots is
potentially saving theseeconomies and allowing the world
economy to continue growingdespite less workforce that is
available to do traditional jobsand potentially new jobs.
Now, the other two aspects thatthey're mentioning that are very
(20:56):
interesting when it comes toglobal economical trends, one of
them is what they're calling theReverse distribution of wealth
that started in the seventies,meaning most of the money is
enriching the top 1% versus thepeople who are actually doing
the work and they're claimingthat access to AI can help
reverse that phenomena becausepeople will be able to do more
on their own, not necessarilydepending on enterprises to
(21:19):
support their livelihood.
And connecting that to thetransition in education that
needs to happen.
A recent report states that 23%of the 2024 Harvard Business
School grads were stillunemployed three months after
graduation.
That's very rare.
Over 4 million members of theGen Zs in the US not employed
(21:40):
and not in school or any kind oftraining joining this new
concept called NEAT NEET, thatstands for not in education,
employment, or training.
These are alarming numbers, andagain, it very clearly shows
that the need for entry jobs isshrinking.
But that being said, the flipside of that is there's a lot of
things that AI cannot do yet,and I AI cannot repair a busted
(22:03):
water line.
It cannot install sonar panels.
It cannot build skyscrapers.
It cannot fix power lines orbuild data centers.
And this may signal a veryinteresting shift in our
education system where more andmore people will go towards
these kind of jobs and where thesalaries are actually growing
dramatically.
(22:24):
I mentioned with you last weekthat there's a shortage of a
hundred thousand electricians tobuild the power plants that will
feed the data centers that arebeing built in the next few
years.
Salaries of elevator techniciansand power plant operators are
now crossing the a hundredthousand dollars a year.
That was not the case a coupleof years ago.
Now combine that with the factthat Gen Zs and millennials are
(22:46):
less inclined to work incorporate jobs, mostly because
what they see in social mediathat is reshaping their
perception when they're seeingpeople younger than them or
their age on YouTube and TikTokthat are presumably making
millions either flipping housesor selling clothing or doing
whatever hassle jobs thatthey're seeing.
Or not.
To mention, Kylie Jennerbuilding a billion dollar empire
on Instagram.
(23:06):
You understand that the drive todo something different, that
they can, a nine to fivecorporate job is definitely
there.
and to add a completelyunrelated to AI aspect to all of
that, the Gen Zs and millennialsare about to inherit a hundred
trillion dollars from babyboomers in the next 10 to 15
years, which is the largestwealth transition in history.
(23:28):
That by itself will haveimplications on how the global
economy works.
So we are in for an interestingride in the next few years, and
I will try to keep you postedand as educated as much as I
can, but you need to think aboutwhat's the implication of that
to you and the business thatyou're running on, that you're
in.
A last deep type topic isTrump's visit to Saudi Arabia,
(23:48):
which ended up with a$600billion investment deal from
Saudi to the us.
This includes investment in ai,defense, energy, and potential
other aspects that is supposedto create 2 million new US jobs
based on the White House factsheet.
Now, if you remember Trump'sintegration, he had all the
(24:09):
who's and who's from SiliconValley and top companies behind
him on stage.
If you thought that was just apart of the inauguration party,
then you are absolutely wrong.
All these people went to Sauditogether with the president,
including Iel Musk and AltmanSan Wong and other people like
that, and they all met with theCrown Prince Mohammad bin
Salman.
And this shows you how thisadministrator works.
(24:32):
They're all about growing the USeconomy and the AI in the eyes
of the current administrationand myself plays a very big role
in that.
Now, my only problem with thiswhole thing, or maybe there's a
few other problems, but thebiggest problem that I have with
this whole thing is that meansthat Saudi will hold significant
components of US AIinfrastructure, whether it's
(24:52):
power generation, data centersand so on.
And I don't know if that's ahealthy thing and I don't know
what are the long-term risks ofthat.
All I can tell you is that whenI compare this to the chip
manufacturing situation rightnow, US made every mistake
possible allowing Taiwan tobecome more or less the world's
sole supplier of advanced chips.
And if we're now gonna haveanother foreign nation hold
(25:15):
components of our entire supplychain for ai, I'm not sure it's
a great investment in the longrun, but time will tell.
And now let's shift to the rapidfire topics.
We're gonna start with a lot ofopen AI news like almost every
week.
They seem to be on fire in thepast few months.
So open AI are still inconversations with Microsoft to
approve their new restructuring,process.
(25:36):
As we mentioned last week, theygave up on converting from a
non-profit to a for-profit.
And what they're actually doingis they're converting their
for-profit component that usedto be an LLC to a public benefit
corporation.
I'm not a hundred percent sure,and it's definitely not my field
of expertise to say what thatmeans, but it means that it is a
for-profit entity that can haveshares for different companies,
(25:58):
meaning investors can haveshares in that entity, but that
also has to balance with itsmission to advance AI for
humanity.
And the nonprofit board this.
PBC is still going to controlthe for-profit arm.
Microsoft hasn't approved thatyet.
They are in conversation onexactly how that's gonna look
like and how many shares in thisnew entity Microsoft is going to
(26:20):
have It has been leaked that inthese conversations, Microsoft
is actually potentially lookingfor less equity, but in return
they want to get access to openAI tech beyond 2030, which is
the current agreement.
This tells you that Microsoft isa huge believer in open AI's
ability to achieve its goals andstill be the leading, or one of
(26:40):
the leading companies in thisfield in the next five years,
because otherwise they obviouslywouldn't want that.
They would want more equity inthe company.
That's despite the hugeinvestment that Microsoft is
making with developing their ownAI capabilities.
Now there are more and morerumors and signals that this is
just a step on the way for openai for an IPO, which makes a lot
of sense to me.
(27:00):
This would be the logical thingfor many different reasons.
A, it will allow them to raisesignificantly more money than
the stupid amount of money theyraised so far, but also will
allow the public to be a part ofthat and get access to more
reporting on what's actuallyhappening in OpenAI as it's
happening.
The other huge investor inOpenAI being SoftBank, that is
now actually a bigger investorthan Microsoft is because
(27:23):
they're committed to invest$30billion versus the only 13.75
billion invested by Microsoft.
So SoftBank, CFO.
Approved that they are givingthe thumbs up to this new
restructuring of OpenAI and thatthey're going to maintain their
$30 billion investment.
So initially there werequestions about that because the
$30 billion investment weredependent on their converting
(27:45):
from a non-profit to afor-profit by the end of this
year.
Otherwise, that number would'vebeen cut by$10 billion to only
$20 billion investment.
But now they've confirmed thatthe full investment is secure
based on the equity they'regoing to get in this new entity.
Now, to show you another angle,if you needed that on, how
strong is open AI's positionacross everything.
(28:06):
BBVA one of the largest banks inthe world is increasing its open
AI agreement from 3000 licensesto 11,000.
ChatGPT Enterprise licensesequipping 10% of their employees
with these AI tools now thebank's initial trials with 3000
licenses saw that chat.
PT saves employees nearly threehours a week every week to every
(28:26):
employee on average, whichobviously drives a lot of
productivity across Teslacoding, customer service, data
analysis, and more.
Something that they are doing inBBVA that I've seen other
companies do, which isincredible, is they've launched
their own chat, GPT store withover a hundred tailored
applications that employees haveaccess to, that were developed
by other employees and theirdevelopment team.
(28:48):
This means that they're now ahundred AI tailored apps and
solutions specifically fordifferent aspects in the bank
that employees have access to.
Now, I wanna expand on touch ofthat point for a minute, and I
know we're not in the deep divesection, but it's very important
for you guys to understandbecause it's a huge opportunity.
The first one is licensesdifferent than Microsoft
copilot, where if you want youremployees to have copilot, every
(29:11):
employee needs a co-pilotlicense because if that's the
way the licensing works, youhave a Microsoft license and you
want to add co-pilot on top ofthat, and nobody can share that
license with open ai, you canhave multiple employees using a
single license.
I don't know if you ever thoughtabout it, but you can have your
entire marketing team using oneCHA PT license, or two or three
depending on how large is theteam.
But you don't necessarily need alicense per person.
(29:32):
The same thing in other aspectsof the business.
This only means that multiplepeople are sharing the same
conversations, but as long asthere's no problem with
information sharing, that wayyou can save yourself a lot of
money, still enjoying thebenefits.
Now, will OpenAI close thatloop?
Maybe at a certain point, butright now it's definitely a
possibility and many companiesthat I know and many of my
clients are doing that rightnow.
(29:53):
The other thing is teaching youremployees how to develop custom
GPT and other AI solutions isthe most effective way to apply
AI quickly, efficiently,effectively across multiple
aspects of the business.
I have weekly meetings withseveral of my clients and in
every meeting a department comesup with a question, idea,
(30:13):
solution, or direction that theywant to take with AI in their
little world.
And together we build what theyneed in order to apply it in
their universe.
In every one of these sessions,it's actually a training
session, meaning I'm not justbuilding it from them, we're
building it together so theyknow how to continue build it
themselves.
And what happens is every singleweek we develop an AI solution
in many cases, as simple as acustom GPT that can save hours
(30:38):
to a specific department everysingle week.
With every single meeting thatis either one hour or two hours,
the companies are gettingsavings of several hours per
week in that one department.
This is a very dramatic shift inefficiency across every aspect
of the business.
And again, I'm doing this withmultiple businesses across all
the departments.
(30:58):
So if you want to enjoy the samekind of thing you need to train
employees in your business onhow to build things that are as
simple as custom gpt or moreadvanced capabilities, and start
enjoying these benefits.
If you wanna learn how you cando it in your business, how can
you train your people?
Just reach out to me on LinkedInor send me an
email@esratmultiply.ai and Iwill gladly share with you the
training programs that we have.
(31:18):
But it doesn't have to be me.
It can be any other source thatcan give you this kind of
training in an effective way.
It is definitely worth theinvestment.
Another big piece of news fromOpenAI.
They just released GPT-4 0.1 tochat GPT.
So GPT-4 0.1 was released a fewweeks ago, but only for the API.
The goal of 4.1 was mostly to bemore competitive in the vibe
(31:40):
coding universe.
And so the API came first, butnow GPT-4 0.1 is available in
ChatGPT in the dropdown menuunder the more models section.
So all the way on the bottom.
You need to remember again thatwhile this is a great model for
coding, it is geared for coding,and it also apparently is very
good at following sophisticatedinstructions.
So if you have a complexprocess, that might be the right
(32:01):
model for you, but just the factthat they're quote unquote,
hiding it under the more modelsis telling you it's probably not
as good as GPT-4 oh or oh threein day-to-day tasks that most of
us are using chat GPT-4.
Another really interesting andpowerful capability from ChatGPT
is the ability to now connectChatGPT Deep research to
(32:21):
OneDrive and SharePoint.
Basically allowing you to queryall the data that is in a
SharePoint site or a SharePointdrive, which is something that
was very desperately needed bymany different companies.
The only solution that providedthat before without going crazy
with dedicated enterpriseinstallations was Perplexity
(32:42):
Enterprise that could do thesame thing, and now that is
available in deep research inpt.
If you click on deep research,there's a little plus button you
can click on that, and then youcan select to connect it to
these tools and then query thoseas well.
This is now available to anypaid customer, so plus pro teams
and enterprise users.
It provides a benefit beyondMicrosoft copilot that is
(33:04):
supposed to provide the samekind of capability, but it's
limited with the amount of filesthat you can actually connect
and query.
There is no clear definition inthe ChatGPT universe on what's
the limitation on how many filesor what the size of the files
that can be in those folders.
I'm sure that we're gonna learnthat in the next few weeks as
people start using it, but thisis an extremely powerful
capability that is now availablein your regular CHA GPT license,
(33:26):
but that's not all CHA GPT DeepResearch can now export PDF
files.
That can also include tables,images, graphs, and so on in A
PDF format, straight out of deepresearch.
So you can now run deep researchon external and internal
information and create a PDFreport that you can then share
with whoever you need to shareit with.
Now, they're also building moreand more integrations for deep
(33:49):
research and Chachi in generalto tools like Dropbox, linear
SharePoint, Datadog, Zendesk,box Resend Teams, SendGrid, gold
Cast, HubSpot Intercom, and manyother tools.
And they're building support forcps, meaning you'll be able to
connect any MCP server out thereto ChatGPT and use that
connection to get informationback and forth, which will make
(34:11):
ChatGPT extremely more powerfulthan it is today.
For business context And nowfrom open AI to their big
frenemy, Microsoft.
Microsoft just announcedslashing 6,000 jobs in the next
step of its restructuring.
That's about 3% of its globalworkforce and it's coming across
multiple aspects of Microsoftworkforce.
(34:33):
It is their largest cut sincethe end of 2023 where they cut
10,000 jobs.
And the quote from the statementfrom Microsoft themselves was,
we continue to implementorganizational changes necessary
to best position the company forsuccess in a dynamic
marketplace.
Now you combine that with thefact that Microsoft pledged to
invest 80 billion in AIinfrastructure in 2025, and with
(34:56):
its pressure to additionalprofitability and you understand
the need to cut jobs.
You combine that with the factthat Satya Nadella said that 30%
of their code is now generatedby ai.
And you understand that theamazing returns that they're
getting right now mostly onAzure Cloud and AI
infrastructure, is notnecessarily fueling more jobs,
but actually allowing them topotentially cut jobs while
(35:17):
maintaining this growth.
The biggest portion of people inthat group are software
engineers and middle managersand.
Out of those 6,000, almost 2000are Redmond alone.
So if you think about aconcentration of a lot of people
in the same field losing theirjobs, it's not gonna be easy to
find jobs, especially when thesame trend of AI replacing these
(35:38):
jobs it's happening in othercompanies as well.
Now they're not the only ones.
So meta cut 5% of their globalworkforce, Salesforce, a
thousand layoffs in 2025.
And a lot of other techcompanies, whether you see it in
the news or not, are becomingmore lean, letting people go and
replacing them with AIcapabilities.
The combination of that with ashaky global economy, with
(35:58):
tariff wars and so on is aperfect storm, hitting the
global and definitely westernjob markets.
Staying on Microsoft, they justadded a hey copilot voice
feature to activate copilot onPC computers.
So if you have that capabilityturned on, you can now say, Hey,
copilot, and just speak withyour own voice to copilot and
ask it to perform differenttasks for you on the computer.
(36:20):
We mentioned that many timesbefore, voice is going to most
likely replace keyboards as themain interface between us and
computers.
I switched almost completely tovoice typing on my computer.
a lot of it is because I startedusing the advanced voice mode in
chat pt, and this just led me tounderstand that I can actually
type that way as well.
And now I use the voice typefeature on the Mac for many of
(36:42):
the things that I'm doing,whether I'm writing documents,
exchanging emails, or writinglong prompts in different AI
tools, voice typing is allowingme to do this significantly
faster.
There are third party tools outthere as well that do an even
better job than the internaltool on the Mac.
The reason I don't use them is Idon't know exactly where the
data goes as I don't knowexactly where the data goes, and
(37:03):
then they track everything thatI'm doing.
And so for now, despite the factit's not the best tool out
there, it's a good enough toolfor me to use, and I use the
Internal Mac tool, by the way,to complete the knowledge about
this new functionality fromMicrosoft.
The.
Wake up word happens locally,but to analyze the information,
it actually requires internetaccess.
So you can only use the heycopilot feature if you have
(37:24):
internet connection.
Another big announcement fromMicrosoft is they announced that
they're going to support A to a,which is agent to agent protocol
that was open sourced by Googlejust a few weeks back, and it's
going to become available onAzure AI Foundry and copilot
Studio in the next few weeks.
There are now two large scaleprotocols that have been
(37:45):
accepted by many companiesaround the world.
MCP is one, which was opensource by Anthropic and MCP
allows agents to connect to datasources and tools in a standard
way, which is taking the worldby store with, I don't even know
how many, but probably millionsof MCP servers available right
now from multiple differenttools and connectors.
And on the other hand, A to a,which is agent to agent, which
(38:05):
is a protocol that allows agentsto talk to other agents in a
standard way, regardless of theplatform or the tools you use to
develop them.
The combination of these twothings.
Is creating a globalinfrastructure in which agents
can collaborate with themselvesand connect to different tools
in a much more efficient way.
That was available just a fewmonths ago, and that is going to
dramatically change everythingwe know about agents and how
(38:28):
they're going to work.
KPMG and other companies basedon the research, are expecting
the agent market to reach$7.8billion this year and$52 billion
by 2030.
I actually think that's lowbowling, where it's gonna be in
2030, because I think agents aregonna take over, more or less
everything that we do.
Meaning 52 billion is a drop inthe ocean compared to what it's
(38:49):
actually going to do.
And from Microsoft to Google,Google DeepMind made a very
interesting announcement thisweek where Alpha Evolve, which
is a new model that they'vedeveloped, can now autonomously
design and optimize complexalgorithms, surpassing human
solutions in math and computingalgorithm creation.
This tool is using Gemini Flashfor rapid idea generation of
(39:11):
what these algorithms can do,and then Gemini Pro to actually
do deep analysis and researchand come up with these new
algorithms.
Google is already seeing hugesuccess with using Alpha Evolve
to create new capabilitiesacross multiple aspects of what
they're doing.
Including metrics, manipulationthat is outperforming deep
minds, alpha tensor chips, theway they're built right now.
(39:33):
So, this is another step in thedirection of AI agents creating
better AI agents if they can, ontheir own, spin up better
algorithms.
That's a very big step towardsthe singularity in which the
acceleration is just not goingto be stoppable because they'll
be able to continuously createbetter and better AI
capabilities on their ownwithout the need for human
(39:54):
input.
I.
Staying on new tools andcapabilities.
Hugging face just released OpenComputer Agent, which is a free
cloud-hosted AI tool thatautonomously performs computer
tasks similar to open AIoperator only with one big
difference.
It is free and it's open source.
It is powered by Quinn vl, whichis a Chinese model with great
(40:14):
visual capabilities, and itknows based on that model to map
everything on your screen and soit knows where to click with a
virtual conceptual mouse andkeyboard and actually can take
over and do every task that youcan imagine.
I think the fact that it's opensource is gonna be very
appealing to companies to comeand integrate into their tools
and so.
This is just another step in theacceleration of AI agents that
(40:35):
can take over tools that we'reusing today, not even building
new stuff that can eliminate thestuff that we're doing today.
They can just use the toolswe're using today in a much more
efficient way and without theneed for us to be there and
actually do the work.
Now it is currently working alittle slow because a lot of
people want to use it, and thepopularity is going through the
roof right now.
I'm sure they will figure it outin giving it the right compute
(40:57):
or the right cost or the rightcost and compute in order to run
this more smoothly.
But even as it is right now, youjust have to wait a few more
minutes for it to do the taskinstead of you doing it.
And so I think it's still worththe wait.
Staying on models and newreleases.
Windsurf, which as we mentionedin the past few weeks, might be
acquired by OpenAI for$3billion, just released their own
(41:17):
in-house family of models thatthey call SWE dash one.
These are frontier level AImodels designed specifically for
software engineering.
They're claiming it matches theperformance of top models from
Claude and GPT, such as ClaudeSauna 3.5 and GPT-4 0.1.
And that it's surpassing them incomplex software engineering
(41:38):
tasks, even beyond just codegeneration.
There are three differentmodels.
SWE Premium for Advanced Tasks,SWE one Light, which is free and
replacing what was calledpreviously Cascade Base and SWE
one Mini, which is also free andis supposed to be quick for
coding predictions.
So three different models forthree slightly different types
(41:59):
of tasks.
All developed by Windsurf anddeployed by them as well.
How will that be integrated withthe new developments from Open
AI and G PT 4.1 and the factthat they might merge into one
company is unclear yet, but Iwill keep you posted as it
evolves.
Speaking of mergers andacquisitions and investments in
the AI world, Databricks, justacquired Database Startup Neon
(42:22):
for$1 billion.
Neon actually has a really coolsolution that allows AI agents
to spin up databases for the useof other AI agents in less than
500 milliseconds.
It basically allows the creationof databases on the fly as
needed by AI agents to performdifferent tasks.
Their recent telemetry data isshowing that over 80% of the
(42:45):
databases that exist on the nearplatform was created by other AI
agents and not humans showingyou the flexibility that this
platform provides, which makesit a very logical acquisition
for Databricks that is focusingalmost solely on AI future of
businesses and how data needs tobe structured and how to deploy
it effectively, and it's justgonna strengthen Databricks
(43:08):
position in this universe.
Now another point on Databricks,they just hired Stefan Orban,
who is the former Google Cloudand AWS executive who has led
big initiatives in both thesecompanies.
He was Google's cloud VP forMigrations and Marketplace, and
AWS VP and general manager ofAWS marketplace.
So he brings decades of cloudcomputing expertise to
(43:32):
Databricks, showing you how muchthis company is going all in on
AI infrastructure and growingtheir market and customer base,
staying on funding routes thatis actually more related to
Windsurf.
Any sphere, which is the companybehind the AI powered coding
tool.
Cursor, secured a 900 millionfounding ground led by Thrive
Capital, Anderson, Horowitz,Xcel, and a few other big names,
(43:54):
valuing the company at$9billion.
Now, to tell you how crazy thisis in December of 2024, so six
months ago, the company raised ahundred million at a$2.6 billion
valuation, so their valuationalmost quadrupled in just six
months.
The combination of that plus thepotential acquisition of
Windsurf shows you how burninghot the AI coating war is right
(44:19):
now.
And I don't think it's gonnastop anytime soon.
And if you connect it back toall the news we talked about in
the beginning as far as peoplegetting jobs, people losing
jobs, Microsoft laying offpeople meta and so on, it's
telling you this is notstopping.
If anything, it's justaccelerating, but it's not
unicorns and butterflies forevery AI company out there,
cohere, which is a smallercompetitor in the field, which
(44:41):
we don't talk about a lot, hasmissed their projections for
2024 by 80%.
He projected$450 millionannualized revenue for 2024, and
it's just on the pace to hit 70million for that pace in
February of 2025.
ISTs are claiming that theirbiggest problem is obviously the
elephant in the room, OpenAIreleasing more and more
capabilities that they justcannot compete with, especially
(45:04):
at the price point that OpenAIis releasing them.
And Cohere has made a strategicchange in the middle of 2024 to
focus specifically on developingenterprise level applications
versus better underlying modelsthat help them to still grow
dramatically between 2024 to2025, but not even close to
hitting the numbers it wassupposed to hit that obviously
(45:24):
has a very significant snowballeffect where investors will
probably be less likely toinvest in them for the next
round.
And in parallel, companies andenterprises may not be willing
to bet on them because of thatsame exact reason.
If they run out of money, theneverything you're building on
the enterprise may go down thedrain together with them.
So sadly, I don't see a verybright future for cohere.
Maybe one of the big playerswill buy them for their existing
(45:46):
enterprise access and that willsave them and the stuff that
they develop so far.
But other than that, it's gonnabe very hard to compete with
OpenAI, Google, and the reallybig players, just because they
have really deep pockets rightnow, access to capital,
knowledge, distribution,compute, and everything else
that is required in order toreally be competitive in this
field.
Another company that announcedan interesting tool that will
(46:08):
help many businesses in AI baseddata analysis is S-A-P-S-A-P.
SAP, just announced what they'recalling Business Data Cloud or
S-A-P-B-D-C, which is amouthful, but the goal is
obviously to leverage all thedifferent data points that are
available in SAP for businesses,which is a lot of data points
from many different sources, andcreate one single data fabric
that enables decision makers tofind relevant information and
(46:31):
get insights from that data.
SAP is claiming that S-A-P-B-D-Creduces data to decision latency
by up to 80% compared to the wayit was done traditionally so
far.
Now, in addition to theinfrastructure itself, they are
developing more and more powerintelligent apps for planning
ai, advanced analytics and soon, that will allow businesses
to benefit from this new unifiedfabric of data.
(46:55):
Now this is a necessity rightnow because a Salesforce survey
reveals a staggering 18% drop inbusiness leaders' confidence in
data-driven decisions comparedto just 2023.
with less than half feelingsecure of their data
reliability.
Think about what I just said.
more than half top executives donot feel that they use to make
(47:18):
significant strategic decisionsis reliable.
So this new world of agenticanalytics, that is just the next
evolution of businessintelligence, is going to
potentially help a lot withthat.
I think will require significantinvestments in data cleaning and
application building and testingand so on.
But I think it will lead to aworld in which any person who
(47:40):
needs to make a decision can geta huge amount of information
summarized at whatever level andformat they need.
And it's gonna be at thefingertips of every single
employee that has access tothese tools.
And that's gonna be a verydramatic change to what most
companies and most seniorexecutives are experiencing
right now.
Which I believe is a very goodthing.
To be fair, you can start doingthis in your business today.
(48:01):
I'm right now with some of myclients are building tools that
knows how to integrate data fromseveral different sources and
provide either a written reportor a spreadsheet or charts and
graphs and dashboards that donot exist in the systems the
company is using right now.
Especially when you're trying tocross data from, let's say
marketing and sales, or salesand customer service and so on.
And now all you need to do is totake CSV exports from these
(48:23):
files, drop them into one place,and get a detailed analysis in a
format that you want that can becreated in.
So within a few hours today, youcan have something like this in
your company that can give yousignificantly better information
with significantly less workthan the way you're doing it
right now.
Another big story this week wasthe backlash in several
(48:43):
different places in the USagainst data centers and their
impact on rural areas.
One story was about X's newsupercomputer center in Memphis
that consumes over 150 megawattsof electricity annually.
That doesn't mean to most of usanything, but it means the same
amount of power that can power ahundred thousand homes that
obviously will strain the local.
(49:06):
The local infrastructure andpotentially can cause
significant environmental harmas well.
Now, while the mayor of Memphisis focusing on the potential of
creating high paying jobs andboost the economy in the area,
it's still not addressing theissue of raising costs of water
and power supply.
'cause it uses a lot of waterfor cooling as well as I
mentioned.
What's the environmental impactof these centers?
(49:28):
Similar concern are soundingacross multiple other places in
the US Including Warren Countyin Virginia.
They had conversations withDominion Energy representative.
That estimate that the datacenter power needs in Virginia
has skyrocketed from 21gigawatts in July of 2023 to 40
by December of this year.
So this is more than two X.
They're also claiming that manyof these data centers are using
(49:50):
diesel generators as backupwhenever they have power surges.
And these pollute and make a lotof noise for nearby
neighborhoods.
And the other concern is thatmany of these data centers got
significantly tax benefits inorder to be built in those
areas, which takes away many ofthe potential revenue that was
supposed to come in, which ispart of the promise of building
(50:11):
data centers in these ruralareas.
One of the city councilrepresentative Eric Gagnon
revealed that a secondChick-fil-A would bring in more
tax revenue and jobs than Amazondata center that only employs 25
contract workers at this point.
So, where is this heading?
I spoke before about potentialgeneral pushback against AI from
(50:33):
people.
I always assumed it's gonna bearound job losses, but
apparently the data center inrural areas might get a bigger
pushback than what I wasthinking.
That being said, there is waytoo much money involved that I
think the rejection by peopleagainst it will actually slow it
down.
but I hope it will, or I hope atleast it will push us to do this
(50:54):
in a more environmentallyfriendly way.
And now just something strangeand interesting that happened in
a courtroom this past week.
In a road rage murder incident,the sister of the victim and her
husband, both with techbackgrounds, created an AI video
of the victim that, again, to beclear, is dead.
That is sharing his thoughts onthe incident, and they showed it
(51:16):
in court.
As part of a testimony.
They were trying to capture thevictim's words and how he would
feel in the current situation,he even forgave the murderer in
the case.
The judge was highly impressedand touched by the video and
increased the sentence ofGabrielle ti Jita who was
(51:36):
convicted for a manslaughter ofChristopher Poey and obviously
the defense in this particularcase have immediately filed for
a, an appeal, basically sayingthat the judge shouldn't be
impacted by this video becausewe don't actually know what this
person would've said.
These are just concepts that areintroduced in a very
sophisticated way using AI as ifthe person has actually said
(51:57):
them.
Now, well, there's obviously twosides to that story and I'm not
sure which one makes more senseto me at this point.
It definitely shows where thefuture that we're going to, it
is going to be very hard to knowwhat is real and what is not.
Whether the person in thisparticular case, it's very easy
to know because the person isdead, but the same exact thing
could be done for live peoplepresenting specific points of
(52:18):
views and things that theypresumably said without being
able to distinguish whether itwas the actual person or not.
I see that as a very seriousissue in this new AI era we live
in, and I think the biggestissue is that most people don't
know that's possible.
That's it for this news episode.
We will be back on Tuesdayshowing you a detailed
comparison between cha, midJourney and Gemini image
(52:39):
generators for business ad andpromotional product creation.
It's a fascinating episode,including showing all the
different prompts and everythingthat you need to know in order
to create amazing visuals foryour business and showing you
which tools you want to use fordifferent use cases.
Don't forget to share thispodcast.
If you like it, just open yourapp right now.
Click on the share button andshare it with several people
(53:00):
that can benefit from it aswell.
And while you're at it, I wouldappreciate if you do two things.
One is fill up the survey.
There's a link to a survey ofwhat you think about this
podcast, what you like, what youdon't like, and will allow us to
improve the podcast to betterfit your needs.
And also you can rank us andgive us a review on your
favorite podcasting platform.
I would really appreciate that.
You can also sign up to ournewsletter and join our Friday
(53:22):
Hangouts and be a part of ourLively AI community.
And you can do all of those withlinks in the show notes.
That's it for today.
Have an amazing rest of yourweekend.
Hello and welcome to a Weekend News episode of the
Leveraging AI Podcast, thepodcast that shares practical,
ethical ways to improveefficiency, grow your business,
and advance your career.
This is Isar Meitis, your host,and we are going to cover some
very profound and interestingtopics that happened this past
week.
The first one, we're going totalk about the first signs of
the end of the internet as weknow it.
(02:31):
Yes, I know that sounds big, butthat's the reality.
The second topic that we'regoing to talk about is how
impactful is AI in highereducation right now, and how it
is going to change highereducation and the workforce in
the next few years, which isgoing to impact all of us.
And the third main topic isgoing to be Trump's visit in
(02:52):
Saudi and how that is connectedto AI and AI infrastructure.
But before we begin, there's twothings I want to add.
First of all, next week I'mgoing to do a special deep dive
in the first section of theepisode, instead of the regular
weekly things I'm going to diveinto the AI Company of the
Future.
It's a podcast that was releasedby dsh who has an amazing
podcast that you should followif you're not following already.
(03:13):
he is a brilliant mind and he'sa great interviewer and he has
access to anybody he wants inthe AI world.
But he was doing actually a soloepisode talking about the AI
company of the future.
You can go and listen yourself,but I'm gonna cover it, in depth
next week.
I wanted to do it this week, butthere was too many important
things to talk about, which Ihope I'll be able to avoid, next
week and jump into that.
And the last thing I wanna saybefore we start is I wanna say
(03:35):
thank you.
so first of all, thank you toall of you for listening to this
regularly, but it's amazing tome how close this community is
becoming of people followingthis podcast and engaging with
me across multiple channels.
We have now several differentweeks, more than 20 people every
week on the AI Friday Hangouts,which you are more than welcome
to join, we'll do this at everyFriday at 1:00 PM Eastern.
(03:55):
If you wanna join, there's alink to do that in the show
notes.
But I want to give a specialthing to Charlie Hoing, Leon
Crisp and Dennis Kovich.
All three of them heard anepisode I released a couple of
weeks ago saying that I like toget an access to Manus and I
don't have access yet.
And all three of them reachedout and offered an invitation
from them to give me access tothat, which I now have and can
(04:17):
experiment and try differentthings, which I will report in
an episode in the next couple ofweeks.
So thank you all of you forlistening.
Thanks specifically to Charlie,Leon and Dennis for caring and
for offering me Manusinvitations.
And now let's get started withthe episode.
(04:37):
So I mentioned the first topicis the end of the internet as we
know it.
I talked about this in manyprevious episodes on how AI is
changing the way we engage withinternet, and how agents and
search are gonna be impacted bythat, and how my personal
behavior has changeddramatically, where probably 85%
of what I'm looking for is notcoming from going to Google, but
(04:58):
going through different AItools.
But this week there was a crazeon Reddit on the data about the
traffic to the top 10 leadingwebsites in the world.
This is the first time in maybethe history of the internet, but
definitely in recent years thatthe traffic to nine out of the
top 10 websites has decreasedmonth over month, and there's
(05:20):
only one website that saw anincrease in the top 10, and that
is ChatGPT.
That, by the way, was number 15on the list just a year ago, and
now it is in number five andgrowing.
What does that tell us?
Well, first of all, I need tosay that this is not one data
point, but the one data point isvery significant.
The regular traffic to websitesis declining.
(05:41):
Apple also reported the overalltraffic to websites and browsers
is declining with more and moreusers globally are now using
ChatGPT Gemini, and other appsto research things that they're
trying to find.
This, as I mentioned again inprevious episodes as well, is
going to have profoundimplications on every online
(06:03):
presence of any company.
If the future of your businessdepends on website traffic as it
looks today, even if it's 30% ofthe revenue that you're making,
40% of the revenue that you'remaking, if you're not going to
make dramatic changes in thenext few years, that is going to
go away because the internet isgoing to be browsed through
agents.
You will talk to whateverapplications, whether it's
(06:25):
ChatGPT, Gemini, Claude Grok, oranything else that will come out
in the next few years, and thesewill do the research for you and
we'll get you back information.
And the companies that willunderstand how to build the
backend of their website in away that will be agent friendly
will win.
And anybody who is late to thatgame will lose potentially their
entire business.
(06:46):
I'm having conversations withvery senior people in really
large companies, in the billionsof dollars who are not aware of
what I just said, and they'renot planning yet.
That tells me two things that,one, the understanding that
people at the leading edge of AIhas right now that is very
obvious to them is not even onthe radar of most companies in
(07:06):
the world.
And the other thing is thatthere's a huge risk and
opportunity of making the rightmoves to understand how to
engage with agents.
To be fair, it's not awell-defined path,.Meaning it's
not very clear how that's gonnaevolve and how it's gonna look
like in two to three years.
But what is very obvious to meis that it's not going to be
what we see and know right now,or what we see and know and
(07:27):
everything we've learned fromrunning web-based businesses in
the past 20 years.
Now, I said that was just onedata point, but the second data
point comes from an interview byEd DQ from Apple, who revealed
potential plans to shift Safarisearch from Apple to AI tools
like Open ai, perplexity,Anthropic, and so on.
That one sentence sent Alphabetsshares down 7%.
(07:52):
That's$150 billion in marketvalue that were eliminated
because of that.
Q'S specific quote was, AIservices will eventually replace
standard search engines such asGoogle.
Now he's Apple's vice presidentof services..
He doesn't just say things.
I think he was very obvious tohim what's gonna be the
implications of what he'ssaying.
And the other piece ofinformation is that Safari
(08:13):
searches fell for the first timein April of 2025.
So again, another data pointthat shows you where is the
direction going, A, from thevolume of search to traditional
search, and B, from the plans ofone of the largest companies in
the world when it comes togiving people access to data,
meaning Apple in all of itsdevices.
(08:34):
combine that with the US JusticeDepartment lawsuit that is
trying to break up Google andpotentially disallow Apple to
continue using Google as itsprimary engine, as well as
potentially forcing Google tosell Chrome.
And you see that things aregoing to change potentially
faster than we initiallythought.
Now if that's not a big enoughblow to your brain to think
(08:54):
about.
I wanna add a few more points,and I mentioned those in
previous episodes, but I wannatouch on them again because I
think they're critical and maybeyou didn't listen to all the
episodes.
the first one is, what happensto the financial engine of the
internet when, and if thishappens.
Meaning, the first one is whatis going to happen to the
internet from a financialperspective when this transition
happens?
(09:14):
So what I mean by that is, thinkabout it.
The internet is the mostincredible service, potentially
the human kind ever had.
That is free.
None of us is paying for theinternet, mean we are paying the
ISP to get internet access atour house.
But we're not paying for theinternet itself.
And the reason for that isbecause ads are paying for the
internet.
But if I'm using ChatGPT orGoogle or Gemini or Grok or
(09:36):
Manus or whatever the case maybe to access the internet, it's
not seeing ads.
I don't see the ads.
I just see the answers that Iwas looking for.
It's gonna book the tickets forme that I needed.
It's gonna order the food that Iwant and so on.
And that means there are no ads.
No ads, no money, no money whopays for the internet.
And I don't think anybody hasanswer to that yet.
(09:57):
The second question is.
Why would people create newcontent if there is no
incentive?
Meaning the incentive to createnew content that gave us
information was I create qualitycontent that's gonna get found
by Google, that's gonna sendpeople to my website, that will
connect with me, my brand, mycompany, and will buy our
products and services.
That model is going to change aswell, which means the incentive
(10:20):
to create new valuable contentmay or may not be there,
depending how this evolves.
So the question becomes who orwhat is going to create new
content in the future?
And that's another question Idon't think anybody has answer
to.
A lot of it is gonna be AIcreated, but even then you're
asking, okay, what kind ofcontent will it create for what
purposes, based on whatinformation and so on.
(10:40):
And the third question, which ismaybe the most important
question is how do you in yourbusiness, in your strategy, are
preparing for this future whenthe facts of how it's going to
evolve are unclear?
And I think the answer for thatis you gotta stay.
And I think the answer for thatis you gotta start researching
how it's gonna impact yourindustry, your business, your
(11:01):
universe, and in parallel, be ontop of what is changing in the
agent world to slowly try tofigure out and try to make the
changes faster than yourcompetition.
So you can capture market shareas this shift happens.
It's not a shift that's gonnahappen overnight, it's gonna be
a process.
And this process may take 18months to five years, but during
(11:22):
that timeframe, there's gonna bemore and more traffic that is
gonna be agent based versushuman based.
the second topic that we'regoing to dive into is the impact
of AI on higher education.
A very long and detailed articlein the New York Times has shared
findings on research thatthey've done on the impact of AI
in higher education.
And what is happening in highereducation right now is the
(11:44):
younger generation that is inuniversities right now in that
is in colleges and universitiesin the US right now is using
ChatGPT and other AI tools inmore and more ways with more and
more students claiming thatthey're generating 70, 80, 90%
of their work with ai, with themjust adding 10 to 20% of the
(12:05):
effort in order to get, to gothrough, in order to go through
their TA in order to completetheir, in order to complete the
university.
In order to complete the taskthat their professors are giving
them.
Now, this has many differentimpacts, both on students and
professors.
First of all, the students'integrity is at question, but it
(12:28):
just depends on what the rulesare in each university or in
each professor and each roomwith one of them defines it
differently.
But it's very, very clear thatmore and more students are using
AI tools in order to do the workfor them and not disclosing it
as they're doing that.
And yes, there are attempts bydifferent universities to have a
more holistic approach to thisphenomenon, but it's definitely
(12:49):
not the norm and it's definitelynot a standard with every
university and sometimes everyprofessor doing their own thing.
Now to tell you how crazy thesituation is, one of the
examples is a philosophyprofessor at the University of
Arkansas.
,Little Rock caught studentsusing AI to write personal
introductions for an ethics andtechnology class.
So I will let that sink for aminute.
(13:10):
It's a personal introduction.
You're telling about yourself toother student in an ethics
class, and there were studentsusing AI to write that piece of
information.
Now, that's funny, but that'salso really, really scary.
It is scary because what kind ofcritical thinking capabilities
are these students developing?
If every task that they need tocomplete is actually done by ai.
(13:31):
Now it's scary year.
If you think about that, theseare the people that are going to
come into the workforce twoyears from now, because in the
past two years, every task thatthey had, they've done with ai.
How are they going to respondwhen they get into a company in
which they cannot use AI foranything?
Well, they may not know how todo projects without AI because
that's everything they've donefor their four years in college
(13:52):
and university that was supposedto prepare them for working in
the companies.
Will they come and work incompanies that do not have very
open AI policies?
What kind of policies you needto have in place as a company
once you start receiving thesepeople into the workforce, that
their solution for everything isChatGPT or any other AI tool.
Now, professors are reallyfrustrated with the whole
(14:14):
situation again because theydon't have a good answer to how
to address the currentsituation.
In many cases, the universitiesare asking them to grade work of
students that was not done bythe students themselves, but was
done by ai.
It's very obvious to them thatthat is the situation, and yet
they still have to grade itbecause the university currently
does not have a better solution.
And yes, there are some AIcheckers, but it's almost as
(14:36):
accurate as flipping a coin,meaning whether you find that
something was done by AI or not,with these checkers, you still
don't know for sure.
And then what are you going todo?
You're gonna throw somebody offtheir university degree when
it's a 50% chance, 70% chance,90% chance.
Where do you draw the line inthe sand?
If you're saying, I'm gonna messwith somebody's future when I'm
not a hundred percent sure thatthey're actually cheating.
The flip side of that is.
(14:57):
Is it cheating?
Are we really preparing them forthe future if we're not allowing
them to use ai?
So I think this is on one hand,a really scary article that
raises a lot of concerns andquestions.
But on the other hand, it isvery obvious that it's the
biggest opportunity theeducation system as a whole ever
had.
because if you think about howwe are teaching students right
(15:17):
now, whether in elementaryschool, middle school, high
school, or higher education.
It is very, very similar to howwe've been teaching for hundreds
of years.
I live in Florida, not too farfrom me.
There's a beautiful town calledSt.
Augustine.
St.
Augustine has the first woodenschool in the United States.
It was built sometime between1702 and 1716, so that's a very
(15:38):
long time ago.
Over 300 years.
And you know what it has aclassroom with desks and chairs
and a blackboard and a place fora teacher to stand in front of
the students.
That looks and feels very muchlike what we're doing today.
So we have an opportunity todeliver personalized, high
quality training and educationto every kid with the teacher
(15:59):
and or professor being more of amentor that can provide the
human side, the mentorship, thehandholding, the moral and
ethical support for studentsversus the actual teaching
itself, which I believe AI canpersonalize and do better than
what a teacher or a professorcan do in front of a class of 30
to 400 students, depending onthe size of the class in the
specific facility.
(16:19):
How do we then balance betweencritical thinking, AI tools and
other tools that will come up inthe future?
I don't know, but I think itrequires and hopefully will
force a complete revolution ofour education system.
And it doesn't end withstudents.
The problem with professors isas profound, meaning professors
are stuck in this weirdsituation when they themselves
(16:40):
are using more and more AIbecause it's helps them create
better lessons, faster andcheaper, faster and easier.
And on the other hand, they'renot allowing their students to
use it.
But an interesting example camethis week from Ella Stapleton, a
2025 nor Northeastern Universitygraduate, who's demanding the
university to pay her an$8,000tuition refund after she
(17:00):
discovered that her businessprofessor Rick, Arrow would re
arrow would use ChatGPT togenerate lectures and lecture
notes without disclosing it.
So on the student side, there'sthe same level of frustration,
right?
Paying all this amount of moneyin many cases with huge student
loans.
And now only to find out that myprofessor, instead of providing
me the information that helearned through the years, he's
(17:21):
actually using ChatGPT toprovide me information that I
could have gotten for free.
And this obviously doesn't endin the universities because it
immediately reflects oninternships and entry level
jobs.
In a recent report by the WorldEconomic Forum, it was found
that AI automation is takingover entry level jobs and it
reducing traditional internshipson opportunities in a very
(17:43):
dramatic way.
Several different professorsfrom known universities around
the world in that research areemphasizing the role of
preparing students for careersusing internships, but they're
claiming that it's becomingharder and harder for them to
find internships for students inspecific areas.
This just amplifies the impactof the education problem itself.
(18:04):
Meaning, on one hand, thestudents are going to do more
and more of their work with AIwithout actually developing some
of the skills they need for thejob force.
On the other hand, they can'teven learn that on the job as
part of an internship becauseinternships are becoming more
and more rare because companiesare saying, I don't need an
intern.
I can actually do this with ai.
And that obviously stops theentry for the waterfall effect,
(18:24):
that then needs to become moresenior employees in multiple
companies.
I asked that question multipletimes.
If you don't need juniordevelopers, how do you get
senior developers?
How do you get team leads?
How do you get CTOs after that?
And the same question happens inlaw firms, et cetera, et cetera.
These are a lot of questionsthat I don't think anybody has
answers to at this point.
Now, the flip side that I do seeis I see a big opportunities for
(18:45):
companies to be able to shift toAI usage because they will hire
people straight out ofuniversity that have significant
access and experience to usingAI tools across multiple aspects
of things that they had to doduring the university.
So while you have a lot ofemployees who are used to doing
things the way they're used todoing it, you may have a breath
of fresh air when it comes tohiring people out of university
(19:08):
who has been using AI everysingle day for four years, which
is probably not the case forpeople in your business.
So there is a light at the endof the tunnel.
I just don't know how long thetunnel is, and I am not sure how
long it will take us to alladjust to this new situation
because it's shifting very, veryfast.
And the bigger systems such asthe education system, the
medical system, the legal systemand so on, are very slow to act
(19:32):
and it will be a turbulent timethrough this transition period.
Now, on the same topic of AIImpact on the Future economy, a
very interesting article on FastCompany addressed actually two
aspects of a) what they call the100 trillion power shift.
And they're saying two verypowerful things are going to
happen in the global economy inthe next 20 years.
(19:52):
One is the rise of AI that, as Imentioned, they're claiming can
drive a hundred trillion dollarsrelocation in global wealth by
2043.
Basically just short of 20 yearsdoubling the global economy,
fueled by automation andinnovation, what they're
arguing, which we've heard timeand time again that AI is not
just going to take jobs, butit's going to actually help
(20:14):
offsetting the shrinkingworkforces in many countries
around the world, especially inaging nations like Japan.
So I highly recommend, if youhaven't read the book, the End
of the World is just thebeginning that you go and read
that book, or at least listen toit if you're more of an
audiobook kind of person.
It's a fascinating book thattalks in addition to a lot of
other things about the shrinkingworkforce in several different
(20:35):
places around the world and theimplications it will have on
these particular economies.
So being able to offset thatwith AI and then robots is
potentially saving theseeconomies and allowing the world
economy to continue growingdespite less workforce that is
available to do traditional jobsand potentially new jobs.
Now, the other two aspects thatthey're mentioning that are very
(20:56):
interesting when it comes toglobal economical trends, one of
them is what they're calling theReverse distribution of wealth
that started in the seventies,meaning most of the money is
enriching the top 1% versus thepeople who are actually doing
the work and they're claimingthat access to AI can help
reverse that phenomena becausepeople will be able to do more
on their own, not necessarilydepending on enterprises to
(21:19):
support their livelihood.
And connecting that to thetransition in education that
needs to happen.
A recent report states that 23%of the 2024 Harvard Business
School grads were stillunemployed three months after
graduation.
That's very rare.
Over 4 million members of theGen Zs in the US not employed
(21:40):
and not in school or any kind oftraining joining this new
concept called NEAT NEET, thatstands for not in education,
employment, or training.
These are alarming numbers, andagain, it very clearly shows
that the need for entry jobs isshrinking.
But that being said, the flipside of that is there's a lot of
things that AI cannot do yet,and I AI cannot repair a busted
(22:03):
water line.
It cannot install sonar panels.
It cannot build skyscrapers.
It cannot fix power lines orbuild data centers.
And this may signal a veryinteresting shift in our
education system where more andmore people will go towards
these kind of jobs and where thesalaries are actually growing
dramatically.
(22:24):
I mentioned with you last weekthat there's a shortage of a
hundred thousand electricians tobuild the power plants that will
feed the data centers that arebeing built in the next few
years.
Salaries of elevator techniciansand power plant operators are
now crossing the a hundredthousand dollars a year.
That was not the case a coupleof years ago.
Now combine that with the factthat Gen Zs and millennials are
(22:46):
less inclined to work incorporate jobs, mostly because
what they see in social mediathat is reshaping their
perception when they're seeingpeople younger than them or
their age on YouTube and TikTokthat are presumably making
millions either flipping housesor selling clothing or doing
whatever hassle jobs thatthey're seeing.
Or not.
To mention, Kylie Jennerbuilding a billion dollar empire
on Instagram.
(23:06):
You understand that the drive todo something different, that
they can, a nine to fivecorporate job is definitely
there.
and to add a completelyunrelated to AI aspect to all of
that, the Gen Zs and millennialsare about to inherit a hundred
trillion dollars from babyboomers in the next 10 to 15
years, which is the largestwealth transition in history.
(23:28):
That by itself will haveimplications on how the global
economy works.
So we are in for an interestingride in the next few years, and
I will try to keep you postedand as educated as much as I
can, but you need to think aboutwhat's the implication of that
to you and the business thatyou're running on, that you're
in.
A last deep type topic isTrump's visit to Saudi Arabia,
(23:48):
which ended up with a$600billion investment deal from
Saudi to the us.
This includes investment in ai,defense, energy, and potential
other aspects that is supposedto create 2 million new US jobs
based on the White House factsheet.
Now, if you remember Trump'sintegration, he had all the
(24:09):
who's and who's from SiliconValley and top companies behind
him on stage.
If you thought that was just apart of the inauguration party,
then you are absolutely wrong.
All these people went to Sauditogether with the president,
including Iel Musk and AltmanSan Wong and other people like
that, and they all met with theCrown Prince Mohammad bin
Salman.
And this shows you how thisadministrator works.
(24:32):
They're all about growing the USeconomy and the AI in the eyes
of the current administrationand myself plays a very big role
in that.
Now, my only problem with thiswhole thing, or maybe there's a
few other problems, but thebiggest problem that I have with
this whole thing is that meansthat Saudi will hold significant
components of US AIinfrastructure, whether it's
(24:52):
power generation, data centersand so on.
And I don't know if that's ahealthy thing and I don't know
what are the long-term risks ofthat.
All I can tell you is that whenI compare this to the chip
manufacturing situation rightnow, US made every mistake
possible allowing Taiwan tobecome more or less the world's
sole supplier of advanced chips.
And if we're now gonna haveanother foreign nation hold
(25:15):
components of our entire supplychain for ai, I'm not sure it's
a great investment in the longrun, but time will tell.
And now let's shift to the rapidfire topics.
We're gonna start with a lot ofopen AI news like almost every
week.
They seem to be on fire in thepast few months.
So open AI are still inconversations with Microsoft to
approve their new restructuring,process.
(25:36):
As we mentioned last week, theygave up on converting from a
non-profit to a for-profit.
And what they're actually doingis they're converting their
for-profit component that usedto be an LLC to a public benefit
corporation.
I'm not a hundred percent sure,and it's definitely not my field
of expertise to say what thatmeans, but it means that it is a
for-profit entity that can haveshares for different companies,
(25:58):
meaning investors can haveshares in that entity, but that
also has to balance with itsmission to advance AI for
humanity.
And the nonprofit board this.
PBC is still going to controlthe for-profit arm.
Microsoft hasn't approved thatyet.
They are in conversation onexactly how that's gonna look
like and how many shares in thisnew entity Microsoft is going to
(26:20):
have It has been leaked that inthese conversations, Microsoft
is actually potentially lookingfor less equity, but in return
they want to get access to openAI tech beyond 2030, which is
the current agreement.
This tells you that Microsoft isa huge believer in open AI's
ability to achieve its goals andstill be the leading, or one of
(26:40):
the leading companies in thisfield in the next five years,
because otherwise they obviouslywouldn't want that.
They would want more equity inthe company.
That's despite the hugeinvestment that Microsoft is
making with developing their ownAI capabilities.
Now there are more and morerumors and signals that this is
just a step on the way for openai for an IPO, which makes a lot
of sense to me.
(27:00):
This would be the logical thingfor many different reasons.
A, it will allow them to raisesignificantly more money than
the stupid amount of money theyraised so far, but also will
allow the public to be a part ofthat and get access to more
reporting on what's actuallyhappening in OpenAI as it's
happening.
The other huge investor inOpenAI being SoftBank, that is
now actually a bigger investorthan Microsoft is because
(27:23):
they're committed to invest$30billion versus the only 13.75
billion invested by Microsoft.
So SoftBank, CFO.
Approved that they are givingthe thumbs up to this new
restructuring of OpenAI and thatthey're going to maintain their
$30 billion investment.
So initially there werequestions about that because the
$30 billion investment weredependent on their converting
(27:45):
from a non-profit to afor-profit by the end of this
year.
Otherwise, that number would'vebeen cut by$10 billion to only
$20 billion investment.
But now they've confirmed thatthe full investment is secure
based on the equity they'regoing to get in this new entity.
Now, to show you another angle,if you needed that on, how
strong is open AI's positionacross everything.
(28:06):
BBVA one of the largest banks inthe world is increasing its open
AI agreement from 3000 licensesto 11,000.
ChatGPT Enterprise licensesequipping 10% of their employees
with these AI tools now thebank's initial trials with 3000
licenses saw that chat.
PT saves employees nearly threehours a week every week to every
(28:26):
employee on average, whichobviously drives a lot of
productivity across Teslacoding, customer service, data
analysis, and more.
Something that they are doing inBBVA that I've seen other
companies do, which isincredible, is they've launched
their own chat, GPT store withover a hundred tailored
applications that employees haveaccess to, that were developed
by other employees and theirdevelopment team.
(28:48):
This means that they're now ahundred AI tailored apps and
solutions specifically fordifferent aspects in the bank
that employees have access to.
Now, I wanna expand on touch ofthat point for a minute, and I
know we're not in the deep divesection, but it's very important
for you guys to understandbecause it's a huge opportunity.
The first one is licensesdifferent than Microsoft
copilot, where if you want youremployees to have copilot, every
(29:11):
employee needs a co-pilotlicense because if that's the
way the licensing works, youhave a Microsoft license and you
want to add co-pilot on top ofthat, and nobody can share that
license with open ai, you canhave multiple employees using a
single license.
I don't know if you ever thoughtabout it, but you can have your
entire marketing team using oneCHA PT license, or two or three
depending on how large is theteam.
But you don't necessarily need alicense per person.
(29:32):
The same thing in other aspectsof the business.
This only means that multiplepeople are sharing the same
conversations, but as long asthere's no problem with
information sharing, that wayyou can save yourself a lot of
money, still enjoying thebenefits.
Now, will OpenAI close thatloop?
Maybe at a certain point, butright now it's definitely a
possibility and many companiesthat I know and many of my
clients are doing that rightnow.
(29:53):
The other thing is teaching youremployees how to develop custom
GPT and other AI solutions isthe most effective way to apply
AI quickly, efficiently,effectively across multiple
aspects of the business.
I have weekly meetings withseveral of my clients and in
every meeting a department comesup with a question, idea,
(30:13):
solution, or direction that theywant to take with AI in their
little world.
And together we build what theyneed in order to apply it in
their universe.
In every one of these sessions,it's actually a training
session, meaning I'm not justbuilding it from them, we're
building it together so theyknow how to continue build it
themselves.
And what happens is every singleweek we develop an AI solution
in many cases, as simple as acustom GPT that can save hours
(30:38):
to a specific department everysingle week.
With every single meeting thatis either one hour or two hours,
the companies are gettingsavings of several hours per
week in that one department.
This is a very dramatic shift inefficiency across every aspect
of the business.
And again, I'm doing this withmultiple businesses across all
the departments.
(30:58):
So if you want to enjoy the samekind of thing you need to train
employees in your business onhow to build things that are as
simple as custom gpt or moreadvanced capabilities, and start
enjoying these benefits.
If you wanna learn how you cando it in your business, how can
you train your people?
Just reach out to me on LinkedInor send me an
email@esratmultiply.ai and Iwill gladly share with you the
training programs that we have.
(31:18):
But it doesn't have to be me.
It can be any other source thatcan give you this kind of
training in an effective way.
It is definitely worth theinvestment.
Another big piece of news fromOpenAI.
They just released GPT-4 0.1 tochat GPT.
So GPT-4 0.1 was released a fewweeks ago, but only for the API.
The goal of 4.1 was mostly to bemore competitive in the vibe
(31:40):
coding universe.
And so the API came first, butnow GPT-4 0.1 is available in
ChatGPT in the dropdown menuunder the more models section.
So all the way on the bottom.
You need to remember again thatwhile this is a great model for
coding, it is geared for coding,and it also apparently is very
good at following sophisticatedinstructions.
So if you have a complexprocess, that might be the right
(32:01):
model for you, but just the factthat they're quote unquote,
hiding it under the more modelsis telling you it's probably not
as good as GPT-4 oh or oh threein day-to-day tasks that most of
us are using chat GPT-4.
Another really interesting andpowerful capability from ChatGPT
is the ability to now connectChatGPT Deep research to
(32:21):
OneDrive and SharePoint.
Basically allowing you to queryall the data that is in a
SharePoint site or a SharePointdrive, which is something that
was very desperately needed bymany different companies.
The only solution that providedthat before without going crazy
with dedicated enterpriseinstallations was Perplexity
(32:42):
Enterprise that could do thesame thing, and now that is
available in deep research inpt.
If you click on deep research,there's a little plus button you
can click on that, and then youcan select to connect it to
these tools and then query thoseas well.
This is now available to anypaid customer, so plus pro teams
and enterprise users.
It provides a benefit beyondMicrosoft copilot that is
(33:04):
supposed to provide the samekind of capability, but it's
limited with the amount of filesthat you can actually connect
and query.
There is no clear definition inthe ChatGPT universe on what's
the limitation on how many filesor what the size of the files
that can be in those folders.
I'm sure that we're gonna learnthat in the next few weeks as
people start using it, but thisis an extremely powerful
capability that is now availablein your regular CHA GPT license,
(33:26):
but that's not all CHA GPT DeepResearch can now export PDF
files.
That can also include tables,images, graphs, and so on in A
PDF format, straight out of deepresearch.
So you can now run deep researchon external and internal
information and create a PDFreport that you can then share
with whoever you need to shareit with.
Now, they're also building moreand more integrations for deep
(33:49):
research and Chachi in generalto tools like Dropbox, linear
SharePoint, Datadog, Zendesk,box Resend Teams, SendGrid, gold
Cast, HubSpot Intercom, and manyother tools.
And they're building support forcps, meaning you'll be able to
connect any MCP server out thereto ChatGPT and use that
connection to get informationback and forth, which will make
(34:11):
ChatGPT extremely more powerfulthan it is today.
For business context And nowfrom open AI to their big
frenemy, Microsoft.
Microsoft just announcedslashing 6,000 jobs in the next
step of its restructuring.
That's about 3% of its globalworkforce and it's coming across
multiple aspects of Microsoftworkforce.
(34:33):
It is their largest cut sincethe end of 2023 where they cut
10,000 jobs.
And the quote from the statementfrom Microsoft themselves was,
we continue to implementorganizational changes necessary
to best position the company forsuccess in a dynamic
marketplace.
Now you combine that with thefact that Microsoft pledged to
invest 80 billion in AIinfrastructure in 2025, and with
(34:56):
its pressure to additionalprofitability and you understand
the need to cut jobs.
You combine that with the factthat Satya Nadella said that 30%
of their code is now generatedby ai.
And you understand that theamazing returns that they're
getting right now mostly onAzure Cloud and AI
infrastructure, is notnecessarily fueling more jobs,
but actually allowing them topotentially cut jobs while
(35:17):
maintaining this growth.
The biggest portion of people inthat group are software
engineers and middle managersand.
Out of those 6,000, almost 2000are Redmond alone.
So if you think about aconcentration of a lot of people
in the same field losing theirjobs, it's not gonna be easy to
find jobs, especially when thesame trend of AI replacing these
(35:38):
jobs it's happening in othercompanies as well.
Now they're not the only ones.
So meta cut 5% of their globalworkforce, Salesforce, a
thousand layoffs in 2025.
And a lot of other techcompanies, whether you see it in
the news or not, are becomingmore lean, letting people go and
replacing them with AIcapabilities.
The combination of that with ashaky global economy, with
(35:58):
tariff wars and so on is aperfect storm, hitting the
global and definitely westernjob markets.
Staying on Microsoft, they justadded a hey copilot voice
feature to activate copilot onPC computers.
So if you have that capabilityturned on, you can now say, Hey,
copilot, and just speak withyour own voice to copilot and
ask it to perform differenttasks for you on the computer.
(36:20):
We mentioned that many timesbefore, voice is going to most
likely replace keyboards as themain interface between us and
computers.
I switched almost completely tovoice typing on my computer.
a lot of it is because I startedusing the advanced voice mode in
chat pt, and this just led me tounderstand that I can actually
type that way as well.
And now I use the voice typefeature on the Mac for many of
(36:42):
the things that I'm doing,whether I'm writing documents,
exchanging emails, or writinglong prompts in different AI
tools, voice typing is allowingme to do this significantly
faster.
There are third party tools outthere as well that do an even
better job than the internaltool on the Mac.
The reason I don't use them is Idon't know exactly where the
data goes as I don't knowexactly where the data goes, and
(37:03):
then they track everything thatI'm doing.
And so for now, despite the factit's not the best tool out
there, it's a good enough toolfor me to use, and I use the
Internal Mac tool, by the way,to complete the knowledge about
this new functionality fromMicrosoft.
The.
Wake up word happens locally,but to analyze the information,
it actually requires internetaccess.
So you can only use the heycopilot feature if you have
(37:24):
internet connection.
Another big announcement fromMicrosoft is they announced that
they're going to support A to a,which is agent to agent protocol
that was open sourced by Googlejust a few weeks back, and it's
going to become available onAzure AI Foundry and copilot
Studio in the next few weeks.
There are now two large scaleprotocols that have been
(37:45):
accepted by many companiesaround the world.
MCP is one, which was opensource by Anthropic and MCP
allows agents to connect to datasources and tools in a standard
way, which is taking the worldby store with, I don't even know
how many, but probably millionsof MCP servers available right
now from multiple differenttools and connectors.
And on the other hand, A to a,which is agent to agent, which
(38:05):
is a protocol that allows agentsto talk to other agents in a
standard way, regardless of theplatform or the tools you use to
develop them.
The combination of these twothings.
Is creating a globalinfrastructure in which agents
can collaborate with themselvesand connect to different tools
in a much more efficient way.
That was available just a fewmonths ago, and that is going to
dramatically change everythingwe know about agents and how
(38:28):
they're going to work.
KPMG and other companies basedon the research, are expecting
the agent market to reach$7.8billion this year and$52 billion
by 2030.
I actually think that's lowbowling, where it's gonna be in
2030, because I think agents aregonna take over, more or less
everything that we do.
Meaning 52 billion is a drop inthe ocean compared to what it's
(38:49):
actually going to do.
And from Microsoft to Google,Google DeepMind made a very
interesting announcement thisweek where Alpha Evolve, which
is a new model that they'vedeveloped, can now autonomously
design and optimize complexalgorithms, surpassing human
solutions in math and computingalgorithm creation.
This tool is using Gemini Flashfor rapid idea generation of
(39:11):
what these algorithms can do,and then Gemini Pro to actually
do deep analysis and researchand come up with these new
algorithms.
Google is already seeing hugesuccess with using Alpha Evolve
to create new capabilitiesacross multiple aspects of what
they're doing.
Including metrics, manipulationthat is outperforming deep
minds, alpha tensor chips, theway they're built right now.
(39:33):
So, this is another step in thedirection of AI agents creating
better AI agents if they can, ontheir own, spin up better
algorithms.
That's a very big step towardsthe singularity in which the
acceleration is just not goingto be stoppable because they'll
be able to continuously createbetter and better AI
capabilities on their ownwithout the need for human
(39:54):
input.
I.
Staying on new tools andcapabilities.
Hugging face just released OpenComputer Agent, which is a free
cloud-hosted AI tool thatautonomously performs computer
tasks similar to open AIoperator only with one big
difference.
It is free and it's open source.
It is powered by Quinn vl, whichis a Chinese model with great
(40:14):
visual capabilities, and itknows based on that model to map
everything on your screen and soit knows where to click with a
virtual conceptual mouse andkeyboard and actually can take
over and do every task that youcan imagine.
I think the fact that it's opensource is gonna be very
appealing to companies to comeand integrate into their tools
and so.
This is just another step in theacceleration of AI agents that
(40:35):
can take over tools that we'reusing today, not even building
new stuff that can eliminate thestuff that we're doing today.
They can just use the toolswe're using today in a much more
efficient way and without theneed for us to be there and
actually do the work.
Now it is currently working alittle slow because a lot of
people want to use it, and thepopularity is going through the
roof right now.
I'm sure they will figure it outin giving it the right compute
(40:57):
or the right cost or the rightcost and compute in order to run
this more smoothly.
But even as it is right now, youjust have to wait a few more
minutes for it to do the taskinstead of you doing it.
And so I think it's still worththe wait.
Staying on models and newreleases.
Windsurf, which as we mentionedin the past few weeks, might be
acquired by OpenAI for$3billion, just released their own
(41:17):
in-house family of models thatthey call SWE dash one.
These are frontier level AImodels designed specifically for
software engineering.
They're claiming it matches theperformance of top models from
Claude and GPT, such as ClaudeSauna 3.5 and GPT-4 0.1.
And that it's surpassing them incomplex software engineering
(41:38):
tasks, even beyond just codegeneration.
There are three differentmodels.
SWE Premium for Advanced Tasks,SWE one Light, which is free and
replacing what was calledpreviously Cascade Base and SWE
one Mini, which is also free andis supposed to be quick for
coding predictions.
So three different models forthree slightly different types
(41:59):
of tasks.
All developed by Windsurf anddeployed by them as well.
How will that be integrated withthe new developments from Open
AI and G PT 4.1 and the factthat they might merge into one
company is unclear yet, but Iwill keep you posted as it
evolves.
Speaking of mergers andacquisitions and investments in
the AI world, Databricks, justacquired Database Startup Neon
(42:22):
for$1 billion.
Neon actually has a really coolsolution that allows AI agents
to spin up databases for the useof other AI agents in less than
500 milliseconds.
It basically allows the creationof databases on the fly as
needed by AI agents to performdifferent tasks.
Their recent telemetry data isshowing that over 80% of the
(42:45):
databases that exist on the nearplatform was created by other AI
agents and not humans showingyou the flexibility that this
platform provides, which makesit a very logical acquisition
for Databricks that is focusingalmost solely on AI future of
businesses and how data needs tobe structured and how to deploy
it effectively, and it's justgonna strengthen Databricks
(43:08):
position in this universe.
Now another point on Databricks,they just hired Stefan Orban,
who is the former Google Cloudand AWS executive who has led
big initiatives in both thesecompanies.
He was Google's cloud VP forMigrations and Marketplace, and
AWS VP and general manager ofAWS marketplace.
So he brings decades of cloudcomputing expertise to
(43:32):
Databricks, showing you how muchthis company is going all in on
AI infrastructure and growingtheir market and customer base,
staying on funding routes thatis actually more related to
Windsurf.
Any sphere, which is the companybehind the AI powered coding
tool.
Cursor, secured a 900 millionfounding ground led by Thrive
Capital, Anderson, Horowitz,Xcel, and a few other big names,
(43:54):
valuing the company at$9billion.
Now, to tell you how crazy thisis in December of 2024, so six
months ago, the company raised ahundred million at a$2.6 billion
valuation, so their valuationalmost quadrupled in just six
months.
The combination of that plus thepotential acquisition of
Windsurf shows you how burninghot the AI coating war is right
(44:19):
now.
And I don't think it's gonnastop anytime soon.
And if you connect it back toall the news we talked about in
the beginning as far as peoplegetting jobs, people losing
jobs, Microsoft laying offpeople meta and so on, it's
telling you this is notstopping.
If anything, it's justaccelerating, but it's not
unicorns and butterflies forevery AI company out there,
cohere, which is a smallercompetitor in the field, which
(44:41):
we don't talk about a lot, hasmissed their projections for
2024 by 80%.
He projected$450 millionannualized revenue for 2024, and
it's just on the pace to hit 70million for that pace in
February of 2025.
ISTs are claiming that theirbiggest problem is obviously the
elephant in the room, OpenAIreleasing more and more
capabilities that they justcannot compete with, especially
(45:04):
at the price point that OpenAIis releasing them.
And Cohere has made a strategicchange in the middle of 2024 to
focus specifically on developingenterprise level applications
versus better underlying modelsthat help them to still grow
dramatically between 2024 to2025, but not even close to
hitting the numbers it wassupposed to hit that obviously
(45:24):
has a very significant snowballeffect where investors will
probably be less likely toinvest in them for the next
round.
And in parallel, companies andenterprises may not be willing
to bet on them because of thatsame exact reason.
If they run out of money, theneverything you're building on
the enterprise may go down thedrain together with them.
So sadly, I don't see a verybright future for cohere.
Maybe one of the big playerswill buy them for their existing
(45:46):
enterprise access and that willsave them and the stuff that
they develop so far.
But other than that, it's gonnabe very hard to compete with
OpenAI, Google, and the reallybig players, just because they
have really deep pockets rightnow, access to capital,
knowledge, distribution,compute, and everything else
that is required in order toreally be competitive in this
field.
Another company that announcedan interesting tool that will
(46:08):
help many businesses in AI baseddata analysis is S-A-P-S-A-P.
SAP, just announced what they'recalling Business Data Cloud or
S-A-P-B-D-C, which is amouthful, but the goal is
obviously to leverage all thedifferent data points that are
available in SAP for businesses,which is a lot of data points
from many different sources, andcreate one single data fabric
that enables decision makers tofind relevant information and
(46:31):
get insights from that data.
SAP is claiming that S-A-P-B-D-Creduces data to decision latency
by up to 80% compared to the wayit was done traditionally so
far.
Now, in addition to theinfrastructure itself, they are
developing more and more powerintelligent apps for planning
ai, advanced analytics and soon, that will allow businesses
to benefit from this new unifiedfabric of data.
(46:55):
Now this is a necessity rightnow because a Salesforce survey
reveals a staggering 18% drop inbusiness leaders' confidence in
data-driven decisions comparedto just 2023.
with less than half feelingsecure of their data
reliability.
Think about what I just said.
more than half top executives donot feel that they use to make
(47:18):
significant strategic decisionsis reliable.
So this new world of agenticanalytics, that is just the next
evolution of businessintelligence, is going to
potentially help a lot withthat.
I think will require significantinvestments in data cleaning and
application building and testingand so on.
But I think it will lead to aworld in which any person who
(47:40):
needs to make a decision can geta huge amount of information
summarized at whatever level andformat they need.
And it's gonna be at thefingertips of every single
employee that has access tothese tools.
And that's gonna be a verydramatic change to what most
companies and most seniorexecutives are experiencing
right now.
Which I believe is a very goodthing.
To be fair, you can start doingthis in your business today.
(48:01):
I'm right now with some of myclients are building tools that
knows how to integrate data fromseveral different sources and
provide either a written reportor a spreadsheet or charts and
graphs and dashboards that donot exist in the systems the
company is using right now.
Especially when you're trying tocross data from, let's say
marketing and sales, or salesand customer service and so on.
And now all you need to do is totake CSV exports from these
(48:23):
files, drop them into one place,and get a detailed analysis in a
format that you want that can becreated in.
So within a few hours today, youcan have something like this in
your company that can give yousignificantly better information
with significantly less workthan the way you're doing it
right now.
Another big story this week wasthe backlash in several
(48:43):
different places in the USagainst data centers and their
impact on rural areas.
One story was about X's newsupercomputer center in Memphis
that consumes over 150 megawattsof electricity annually.
That doesn't mean to most of usanything, but it means the same
amount of power that can power ahundred thousand homes that
obviously will strain the local.
(49:06):
The local infrastructure andpotentially can cause
significant environmental harmas well.
Now, while the mayor of Memphisis focusing on the potential of
creating high paying jobs andboost the economy in the area,
it's still not addressing theissue of raising costs of water
and power supply.
'cause it uses a lot of waterfor cooling as well as I
mentioned.
What's the environmental impactof these centers?
(49:28):
Similar concern are soundingacross multiple other places in
the US Including Warren Countyin Virginia.
They had conversations withDominion Energy representative.
That estimate that the datacenter power needs in Virginia
has skyrocketed from 21gigawatts in July of 2023 to 40
by December of this year.
So this is more than two X.
They're also claiming that manyof these data centers are using
(49:50):
diesel generators as backupwhenever they have power surges.
And these pollute and make a lotof noise for nearby
neighborhoods.
And the other concern is thatmany of these data centers got
significantly tax benefits inorder to be built in those
areas, which takes away many ofthe potential revenue that was
supposed to come in, which ispart of the promise of building
(50:11):
data centers in these ruralareas.
One of the city councilrepresentative Eric Gagnon
revealed that a secondChick-fil-A would bring in more
tax revenue and jobs than Amazondata center that only employs 25
contract workers at this point.
So, where is this heading?
I spoke before about potentialgeneral pushback against AI from
(50:33):
people.
I always assumed it's gonna bearound job losses, but
apparently the data center inrural areas might get a bigger
pushback than what I wasthinking.
That being said, there is waytoo much money involved that I
think the rejection by peopleagainst it will actually slow it
down.
but I hope it will, or I hope atleast it will push us to do this
(50:54):
in a more environmentallyfriendly way.
And now just something strangeand interesting that happened in
a courtroom this past week.
In a road rage murder incident,the sister of the victim and her
husband, both with techbackgrounds, created an AI video
of the victim that, again, to beclear, is dead.
That is sharing his thoughts onthe incident, and they showed it
(51:16):
in court.
As part of a testimony.
They were trying to capture thevictim's words and how he would
feel in the current situation,he even forgave the murderer in
the case.
The judge was highly impressedand touched by the video and
increased the sentence ofGabrielle ti Jita who was
(51:36):
convicted for a manslaughter ofChristopher Poey and obviously
the defense in this particularcase have immediately filed for
a, an appeal, basically sayingthat the judge shouldn't be
impacted by this video becausewe don't actually know what this
person would've said.
These are just concepts that areintroduced in a very
sophisticated way using AI as ifthe person has actually said
(51:57):
them.
Now, well, there's obviously twosides to that story and I'm not
sure which one makes more senseto me at this point.
It definitely shows where thefuture that we're going to, it
is going to be very hard to knowwhat is real and what is not.
Whether the person in thisparticular case, it's very easy
to know because the person isdead, but the same exact thing
could be done for live peoplepresenting specific points of
(52:18):
views and things that theypresumably said without being
able to distinguish whether itwas the actual person or not.
I see that as a very seriousissue in this new AI era we live
in, and I think the biggestissue is that most people don't
know that's possible.
That's it for this news episode.
We will be back on Tuesdayshowing you a detailed
comparison between cha, midJourney and Gemini image
(52:39):
generators for business ad andpromotional product creation.
It's a fascinating episode,including showing all the
different prompts and everythingthat you need to know in order
to create amazing visuals foryour business and showing you
which tools you want to use fordifferent use cases.
Don't forget to share thispodcast.
If you like it, just open yourapp right now.
Click on the share button andshare it with several people
(53:00):
that can benefit from it aswell.
And while you're at it, I wouldappreciate if you do two things.
One is fill up the survey.
There's a link to a survey ofwhat you think about this
podcast, what you like, what youdon't like, and will allow us to
improve the podcast to betterfit your needs.
And also you can rank us andgive us a review on your
favorite podcasting platform.
I would really appreciate that.
You can also sign up to ournewsletter and join our Friday
(53:22):
Hangouts and be a part of ourLively AI community.
And you can do all of those withlinks in the show notes.
That's it for today.
Have an amazing rest of yourweekend.
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.
Ridiculous History
History is beautiful, brutal and, often, ridiculous. Join Ben Bowlin and Noel Brown as they dive into some of the weirdest stories from across the span of human civilization in Ridiculous History, a podcast by iHeartRadio.