February 6, 2023 • 34 mins
Did you know that research done by Sequoia Capital (one of the largest VCs in the world) predicts that ALL COMPUTER CODE and ALL BUSINESS CONTENT will be CREATED BY Ai by the YEAR 2030.
That is only 7 years away. anyone who is starting college right now, and learning any skills, is at risk of graduating 4 years from now with NO open jobs in their field. Not because of an economic downturn, but forever...
Ai is tasking a bigger part in every aspect of our lives and our businesses. While until just a few years ago, only giants like Amazon, Google, IBM and Facebook benefited from the amazing advantages Ai brings, Ai driven tools are now accessible to everyone on almost every aspect of running a business.
While that has been a clear trend for a few years, 2022 was a tipping point. First with the introduction or maturity of many tools, but mostly with the introduction of Chat GPT, Dall-e, Stable Difusion, and similar tools, that took the world by storm.
If you have never used any Ai tools, strap in, and get ready for a hell of a ride as this episode will blow your mind. If you have been using these tools, you would still gain a lot from this episode as we are covering many use cases that are possible today using such tools.
Hi, It's Isar the host of the Business Growth Accelerator Podcast
I am passionate about growing businesses and helping CEOs, business leaders, and entrepreneurs become more successful. I am also passionate about relationship building, community creation for businesses, and value creation through content.
I would love it if you connect with me on LinkedIn. Drop me a DM, and LMK you listened to the podcast, what you think and what topics you would like me to cover 🙏
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Isar Meitis (01:04):
Hello and welcome to the Business Growth
Accelerator.
This is Isar Meitis, your host,and this is a very special
episode.
I've been working on theresearch for this for a while,
and today we're going to talkabout AI, artificial
intelligence, and how it'simpacting and how I think it's
going to impact, the future ofthe world and more specifically
the future of business.
(01:25):
And what's been happening in thepast few years is that AI has
been getting into more and moreaspects of our lives and more
and more aspects of ourbusiness.
And 2022 has been the tippingpoint where things started
exploding, but before we jumpinto 2022, let's figure out what
happened before and talk alittle bit about concepts.
(01:46):
Most of these episodes is gonnabe very practical with use cases
that people are using AI thatyou can use in your business.
But some background I think willbe helpful, especially to people
who has not been reading orfollowing this world.
Artificial intelligence Is apart of computer science that
allow computers machines tolearn in a similar way that
(02:08):
humans learn.
Meaning instead of writing apiece of code that tells the
computer to do something, itactually learns from mistakes or
from feedback that it's beinggiven based on action that it's
taking.
So there's a whole process oftraining the machine for it to
learn, just like you would trainkids or adults in order to learn
(02:28):
new things.
You give it feedback and then itknows what was wrong and what's
better, and you keep on doingthis again and again and again,
and that's how the machinelearns.
Hence, machine learning is theconcepts behind artificial
intelligence.
When we say artificialintelligence, it's a really
broad term and there's many moregranular terms of this that has
(02:49):
to do with different aspects ofwhat it's actually going to do.
Meaning is it's gonna deal withlanguage processing, is it going
to deal with image processing,video processing, data analysis,
and so on.
Each and every one of them is adifferent aspect of artificial
intelligence and machinelearning and people today are
talking about general ai, whichmeans the capability for a
(03:11):
machine to actually be like ahuman, meaning, be able to do
all the things that we do andnot be specialized at one thing.
And there's still a few yearsbefore that will be possible,
but even the capabilities thatexist today are pretty
incredible.
And we'll touch on a lot of themin this episode.
AI has been used even by youmaybe even without knowing about
it for a while.
(03:32):
If you've been shopping onAmazon, which most of us
probably have, and you seerecommendations of other stuff
you might wanna buy, or if yougo to Netflix and you get
recommendations of the nextvideo, or if you go on Google
and you get results.
All of these things are based onartificial intelligence, meaning
it's taking a huge amount ofdata and it's looking at your
personal touchpoints anddifferent data it has on you,
(03:54):
and it's trying to predict basedon that what's gonna be the most
relevant and that will attractyou to take a particular action.
So you've been interacting withAI systems for a very long time.
What happened in the past fewyears is that AI from being kept
only for giants like Microsoftand Netflix, more and more
companies and startups startedreleasing AI driven tools that
(04:18):
allows companies and individualsto use and leverage AI for
different business and personaltasks.
Let's take a few examples.
copy.ai has existed for a while.
It's a platform that allows youto, like the name suggest, write
copy.
So marketing copy, like websitecopy, newsletter copy, social
media, post copy, et cetera.
(04:38):
It allows you to do all thesethings and it'll write the copy
for you based on differentparameters you give to it.
Another example is RYTR.
It is spelled R Y T R and thewebsite is rytr.me and it allows
you to write longer kind ofcopy, like even blog posts and
longer piece of content based onparameters you give it.
(05:00):
If you want to learn more aboutthese kind of tools, I've done a
whole episode dedicated to RYTRand similar platforms.
It's episode 126 of thispodcast, so just go back and
check that one out.
Such tools became more and moreavailable that started doing
more and more unique and coolthings.
The next interesting iterationwas released by a company called
OpenAI, which you're gonna talkabout more later on in the show.
(05:23):
And OpenAI back in January of2021, released a tool called
Dall-E.
Obviously wink, wink to Dali thepainter.
And what it does, it enables youto put in a text, a descriptive
text, and it creates an imagefor you right there on the spot.
So it doesn't find an image onthe internet.
It literally creates it in realtime based on parameters.
And you can use whateverparameters you want.
(05:44):
And when I say parameters, itsounds very techy, but all you
have to do is type what you wantto paint.
So if you'll say, I want you topaint a house on a moon.
It will paint a house on a moon.
But you can be a lot moredescriptive than that and say, I
want you to paint a house on themoon at sunrise in a realistic
way.
It's gonna generate almost likean image, like picture of a
(06:05):
house on a moon.
But you can also say, I want thehouse to be two stories high and
to have yellow windows, and Iwant three cars parked in the
garage, and it will do that aswell.
Or you can go and say, I wantthat whole thing to be painted
in the Renaissance style, and itwill give you that kind of style
to the painting.
So it's really limitless to thetype of data you can give it,
(06:29):
and the more data you can giveit, the more specific it will
become.
And then gives you options andyou can pick one options and
keep on changing that until youget to an outcome that you want.
These kind of tools startedbecoming more and more available
for video processing, for imageprocessing, for written content
creation.
And so on.
(06:49):
These kind of tools existed fora while and more and more of
them started coming out.
But something dramatic changedin November of 2022.
In November of 2022, OpenAIreleased two tools.
One is Dall-E 2, so it's a moreadvanced form of the first
version of this that is just aimage generator based on text,
(07:11):
but the thing that reallychanged everything was ChatGPT.
ChatGPT is based on their chatengine that allows you to chat
with their bot that they've beentraining for a very long time.
And the reason I'm saying itchanged anything is the fact
that you can literally give itany task that has to do with
(07:32):
organizing content, writingcontent, interpreting content,
creating new content in anywritten form you can imagine.
So this could be a blog post,this could be a legal document.
This could be a song.
This could be a poem.
This could be a summary inbullet points.
This could be computer code.
Literally any format you ask itto produce something, it will
(07:52):
produce it in a very highquality.
And in seconds, and at least fornow for free.
And when I say everythingchanged with the launch of
ChatGPT, I just wanna talk alittle bit about adoption.
You probably heard the termChatGPT a million times in the
past two months.
But if we're talking about amillion, let's talk about how
quickly it took differentplatforms to get to a million
users.
(08:13):
Netflix took about 3 1/2 yearsto get to a million active
users.
Spotify took 5 months.
Facebook took 10 months.
So these are the timeframes ofsome of the best, most
successful adoptions of newplatforms in the world that we
live in.
It took ChatGPT 5 days to get toa million active users, and this
(08:35):
has been continuing to grow atthe same crazy pace.
Now there's been chat platformsbefore that.
So what's the big deal aboutChatGPT and why is it taking the
world by a storm?
The first thing is it's a verysimple and easy to use user
interface.
They're using the same conceptthat Google used when they
(08:55):
launched their search, it's justone line.
There's no menus, there's nobuttons, there's nothing you can
do other than write someinstructions in the line and get
a response and chat back andforth with the chat bot.
So it's extremely easy to use.
But the other thing is that it'sreally extremely powerful and
very, very versatile.
It will do really almosteverything you ask it to do on a
(09:18):
very wide range of topics andtasks that you will ask it to
perform for me, like I mentionedbefore.
So what can you do with it?
First of all, you can writecopy.
So if you need a new concept foryour website, you can ask
ChatGPT to write it for you.
Write me three differentvariations of the content I
(09:38):
should put on my website if I amtrying to attract mechanics to
work for my shop.
And it will give you threedifferent options of marketing
content that you can put on yourwebsite.
You can use it to write copy forsocial media posts, you can take
whatever topic you wanna writeabout this week and say, I would
like you to write a post about.
(09:59):
1, 2, 3 and 4 for LinkedIn andyou will create the content to
be adopted in the right lengthand the right formatting to fit
a LinkedIn post.
You can go a step beyond thatand use reference material as
the baseline for the post.
So let's say you wrote a blogpost and you wanna summarize it
to a post on social media.
You can copy and paste thecontent that you've written, the
(10:21):
long form content, paste it intoChatGPT and say, I want a
summary of this as a post onLinkedIn, and it will do it for
you.
You can ask for it as a summaryin bullet points, and it will do
it for you.
And if talking about summaries,it can summarize literally
anything.
If you have a very long documentyou wanna summarize, you can
copy and paste the document intoChatGPT, ask it to summarize for
you in whatever length orwhatever definition you want.
(10:43):
So I want 20 bullet points as asummary of this whole thing, and
it will do it for you in a veryaccurate and concise way.
You can give it a url.
Let's say you want a summary ofa specific blog post.
You can go and copy the url ifthat blog post, drop it into
ChatGPT and ask it to summarizeit for you, and it will.
Now, that sounds a little bitlike cheating, but not really
because we think about back inthe day when we had to read
(11:05):
specific, really long, boringbooks for graduation.
Most people did not really readthe full books.
They read a summary that youcould buy somewhere at a
bookstore that will actuallyallow you to read 60 pages
instead of 300 pages and stillbe able to pass the exam.
So it's kind of the same conceptonly for any kind of content you
want.
The other cool thing about it isthe more information you're
gonna give it, the more it'sgonna use that information in
(11:28):
order to make your content morespecific.
If you use specific keywords, ifyou use specific ideas, if you
mention specific things, it willuse those in the outcome of the
content that it is generating.
So, Creating any kind of writtencontent, whether it's original
from scratch, based on ideas andconcept, whether it's a summary
of something, whether it's withreference to something, becomes
(11:50):
an effort of seconds instead ofsometimes hours or days.
Another thing that he doesreally well is to write pieces
of code, computer code.
So if you're a computerprogrammer or you're learning
how to do that, you can say whatyou want it to create, give it
the language.
So let's say you want a codingin Java to do something, you
will describe that something andyou will create the Java script
(12:12):
for you to use.
In your code in order to do thatspecific thing.
Now can you program a completeplatform?
Not yet.
I'm sure we'll get there, butprogramming or solving specific
problems with code, it knows howto do that extremely well,
extremely efficient, extremelyfast, and very accurate.
Now take it to the day-to-daybusiness world, you can write
(12:32):
the company update with that.
You can say, these are thethings I wanna talk about.
Can you write me, the speech?
You can use it to write otherspeeches that you wanna put on
stages with specific informationthat you feed into it, just to
break it down into the rightformat, the right structure, the
right wording.
Whatever you want help from itto do, it can do for you.
(12:52):
You can use it for marketresearch.
If you want to know somethingabout your industry or a new
industry you want to go into,you can use ChatGPT for that as
well.
You can go in and say, what arethe 10 largest companies in this
and that niche?
And it will give you the list ofthe companies and then you can
go one by one and say who arethe companies that are most
known to use that particularproduct?
(13:14):
What is that product most knownfor?
What kind of marketing languagehas this company used in order
to attract the people that it'sbeen attracting and so on and so
forth?
You can keep on peeling theonion and drilling deeper and
deeper to find more and moreinformation that before was a
lot harder to get to because youhad to run a Google search and
then look at 50 differentarticles and then from that,
(13:36):
find the information.
But instead, This thing is doingit for you and it's doing it in
again, seconds instead of hoursof research that you have to do.
I'll give you one little warningthough.
ChatGPT is not connected to theinternet.
It's an offline platform, andyes, it's been trained based on
a lot of internet data and thenew variation of it, which is
coming out soon.
(13:57):
We'll have a lot more data, butit's not really online.
As an example, it does not knowthat Elon Musk has bought
Twitter, it does not knowthere's a war between Russia and
the Ukraine.
If you're looking for up to dateinformation, you will not get
it.
So it has some disadvantages andlimitations, but on a very big
level of doing research.
It's an incredible, incredibleresearch tool.
(14:17):
And like I said, you can drilldeeper, you can get summaries,
you can do whatever you wantwith just typing the right
questions.
Next thing it does extremelywell is customer support.
There's more and more platformsbeing developed on top of AI
solutions to do customerservice.
Why?
Because customer service is alogical process.
There's something that happened.
There is an X amount ofpotential outcomes.
(14:40):
There's the way to navigatethem, and AI is very, very good
at doing that.
What we'll see in the nearfuture is more and more customer
service processes being takenover by AI versus actual people,
whether it's a live call, chat,whatever the case may be, that
will be taking over, it'salready happening today, right?
So when we call an ivr, whichtoday is really dumb and
(15:00):
annoying and we hate it, but inthe very near future, it will be
just like speaking to a person,but different than today, that
person will know literallyeverything they need to know and
will have access to everypotential outcome, and will
really be able to navigate veryquickly to the potential outcome
based on the set of guidelinesthat it has.
Same concept obviously appliesto sales, so you can take your
(15:21):
company's knowledge on how tosell your product or service
best.
Train an AI model and that useChatGPT as an engine if you
want, behind the scenes.
Then when people use chat onyour website, they can get very
accurate, very sophisticated,very to the point answers to
their questions without actuallyhaving to deploy salespeople to
(15:42):
do these things.
You can also use it for HRrelated stuff in your company,
like planning the next Halloweenparty.
If you go into ChatGPT and say,I have a hundred employees and I
wanna plan a Halloween partythemed for the eighties, what do
you suggest we would do?
And it will give you a list ofeverything you should do, what
drinks you should prepare, whatcostumes people should wear, how
(16:04):
big of a venue you need to rent,and so on and so forth.
It will help you plan the party.
The options are almostlimitless.
Wanna take it up a notch?
Let's look at some other usecases that you can use in your
business in order to do somemore interesting things.
Let's assume for a second youhave a product or a service and
you want to create demo videosof how to use that particular
(16:27):
thing or specific aspect of itand so on.
The way to do this today, if youwant a person involved is to
rent a studio or build a studioin your office and have cameras
and lighting and spend a lot oftime and a lot of money
recording people in front ofcameras that make mistakes and
have to go back and recordagain, and they get tired and
they get confused.
And then you have to get allthat content and edit it so it's
(16:49):
perfect and it looks great, andthen you can use it in whatever
how-to videos that you want.
Well, that is gonna disappearvery, very quickly because today
there are already tools outthere like Synthesia and D-ID
and a bunch of others thatreally enable you to create a
very realistic avatar.
(17:10):
And when I say very realistic,it looks like people, it talks
like people.
You can still tell with some ofthem it's not human.
It's still good enoughdefinitely for any demo of the
product.
And it takes seconds because allyou have to do is type the text
into it choose an avatar to yourliking.
It doesn't look like a cartoon.
It's an actual person.
It looks like a real person.
(17:31):
And this person, because it's asynthetic person, can be in any
age, any ethnicity you choose,any kind of features that you
would like.
You really have endless kind ofoptions and it can speak almost
any language on the planet.
So you copy and paste.
Text into there and it will sayit perfectly without mumbling,
(17:53):
without going, uh, uh, uh, inthe middle.
So you have to cut and recordagain without spending studio
time, without spending editingtime, and you have a perfect
video in seconds.
Now, let's combine these twothings together.
Let's say you want to create apresentation about a topic.
In video to put it on yourwebsite, to put it on social
media, to put it wherever youwanna put it.
(18:15):
Instead of working in front of acamera, instead of recording all
the content, instead of editingall the content, instead of
doing the research for thecontent, you can go to ChatGPT
and literally give it the inputand saying, I want to create a
lecture on these topics withthis information in mind.
Paste in the information youwanted, and the lecture should
be 45 minutes and I need it inEnglish or Spanish or Italian,
(18:38):
and you will generate it foryou.
And then you take thatinformation to one of the
platforms that I mentionedbefore, like Synthesia or D-ID
or one of those Pasteinformation in there.
Choose your avatar and you havethe video ready to go.
A whole game changer in the waycontent is being created.
(18:58):
So far we talked about creatingwritten content, and we talked
about creating video content.
Now let's talk a little bitabout images and how, again,
these things can combinetogether.
As I mentioned earlier, whenChatGPT was released, Dall-E 2
was released, and as thepopularity of these tools
skyrocketed, other platformsthat are AI image generators
(19:19):
skyrocketed with it.
The two biggest ones togetherwith Dall-E are midjourney and
Stable Diffusion.
All of these basically do thesame thing as I mentioned
before, you give it adescription as detailed as you
want of what you want created,and you will create an image for
you.
That's obviously a very cooltoy, but it's.
Also a very impactful andeffective business tool.
(19:41):
Why?
First of all, because right nowmost companies are paying a huge
amount of money to differentstock photos websites in order
to get specific photos that theywant to use in their marketing
on their website, because theyhave to be licensed for the
people who took the pictures.
But now you can create whateverimage you want in whatever
scenario, in whatever lighting,in whatever concept you want on
(20:01):
the fly, which means you don'treally need stock photo, you
don't need to search for stockphoto.
You can actually create theactual images that you want as
needed per the exact scenariothat you want to use them for.
That's one way to use it.
Another way to use it is indocumentation that you're trying
to create.
If you wanna talk about specificpeople in specific scenarios and
you want to use images thatdescribe that, you can do that
(20:25):
and create those images rightthere and then, instead of going
and searching through hundredsof photos in order to find the
one that you want.
Another interesting and coolimplementation that people
already started doing is peoplestarted creating kids books in
minutes.
And how do they do that?
So you go to ChatGPT and youwrite I wanna write a kid's book
(20:46):
about two brothers and a sister,and you can give them your
specific names that are learninghow to be green and how to
recycle and how to make theworld a better place and stop
global warming.
And I want it to be 10 pageslong, and it will create the
text for you in rhymes accordingto all the parameters you said.
And now for each of those pages,for each section of rhymes about
(21:08):
a specific topic, you go to oneof those image generators and
you describe what the text says,and it will create images for
you.
And you can say, these are forkids' books, and they will be
book friendly kind of drawings.
Now all you have to do is open aPDF editor, place a page from
the written content in front ofit on the other side, place the
(21:29):
image that you just created.
Save the pdf, upload it toAmazon, and you have a kid's
book on a specific topic thattook you minutes to generate.
Now, this may not be your thing,but it gives you an idea of what
use cases you can use this forand how extremely efficient the
process becomes.
The next aspect of images isobviously graphic design.
(21:51):
People use graphic design formultiple different things.
From designing websites, todesigning logos, To designing
brochures and so on and soforth, designing products, and
so far the people who did thiswere graphic designers who got
trained and got experienced andspent years doing so.
Today almost every person can dothese things.
To be fair, most of them are notgreat in graphic design.
(22:14):
Based on most of everything thatI'm reading right now and
researching right now, design isprobably the first complete
revolution of the world whereliterally anybody will be able
to create incredible designswith a few sentences, and then
changing it around until theylike them.
It will be a much faster, muchmore efficient, and a way
cheaper process than it is todaywith graphic designers.
(22:35):
The predictions are that thiswill be really good and really
accessible this year, meaningstill in 2023.
So let's do a quick recap.
This whole concept that I talkedabout so far, it's called
generative ai.
Generative AI allows you to likeits name, so just generate new
things using ai.
This could be video, this couldbe audio, this could be written
content.
And Sequoia recently released aresearch that they've done and
(22:59):
their predictions based on thatresearch says that most likely
100% of the content that we'llconsume by the year of 2030 will
be created by AI versus actualpeople, roughly at the same
time, potentially before that,the research says all computer
code will be AI driven and notwritten by people.
(23:20):
As I mentioned, the sameresearch says that by the end of
2023, graphic designcapabilities will be strong
enough.
It still won't replaceeverything, but it will be
capable enough to replace everygraphic design need.
Now, so far, I talked aboutstuff that is very, very
specific.
It's content creation ofdifferent kinds, right?
So people who are writers,people are technical writers,
(23:41):
people are graphic designersand, and so video producers and
so on.
So you may think, okay, whatmight be protected from this
change?
Probably stuff that is veryhuman, high touch, emotional
kind of thing, I got anothersurprise for you.
A guy named Dr.
Robert Morris, who has a PhDfrom MIT.
(24:02):
He's the founder of Koko.
Koko is an an online emotionalsupport platform.
So it's a chat platform thatenables people, mostly young
adults that have emotionalissues.
To have a chat with somebody whowill support them through their
difficulties, it's an incredibleplatform.
He ran a test with over 4,000people who approached his online
(24:23):
service with more than 30,000messages.
And the way he ran the research,he had a person, professional
human therapist, the people healways uses, but instead of
having them respond to theparticipants, he copied the
participant's question orsentence or whatever they wanted
into ChatGPT and copied andpaste it back the respond to his
(24:47):
participants, who, again, werelooking for emotional support
through chat.
The chat messages get scored,and at the end of each chat
people say how good they feelabout the responses they got and
how much it helped them.
And the ChatGPT messages scoredsignificantly higher than the
human therapists.
In addition to that, the respondtime dropped by 50% and that's
(25:09):
while there was a human in theloop, just to making sure that
Chat PGT doesn't send backsomething that is actually bad
or inappropriate or couldactually harm the person on the
other end.
So if you think about runningthis at scale, without the human
in the loop, you can get toextremely fast respond time.
That is still extremely helpfulthat people that are looking for
(25:30):
support find more helpful thanprofessional human therapists.
It's hard for me to think of ascenario that is requiring more
emotional empathy than thisparticular scenario.
So even this is not protected.
By the way, once hisparticipants learned that
there's a machine in the loop,which they asked them if they
(25:52):
would use it, they would say, ifit was a machine, we would not
use the platform, even thoughthat the vast majority said that
the responses they got wereactually better than the ones
that the human gave them.
Again, they did not know whatwas human and what was Chat PGT
producing.
But it's just coming to show youthe power of this platform.
This may sound like the end ofthe world, right?
More or less, every professionwe know is gonna go away because
(26:13):
the robots are going to replaceit.
The reality is very different.
And first of all, let's go backand look at history and probably
the closest thing to this is theindustrial revolution.
But it's a different scenariothan the Industrial Revolution
because the IndustrialRevolution took years and years
to happen and take over theworld, and this will take
between months and a couple ofyears to do a lot of things
(26:35):
we're doing today.
So the speed in which it'shappening is way faster than the
Industrial Revolution.
The other big difference isbecause so many platforms are
developed on top of AI solutionstoday, while the Industrial
Revolution touched on specificprofessions, this is gonna touch
more or less anyone on theplanet, regardless of what
(26:55):
you're doing today.
some professions faster thanother, obviously, as I mentioned
earlier.
But as I mentioned, theseplatforms have to be trained,
meaning there has to be somebodytraining those platforms,
helping them get better andbetter and better so they can do
all these tasks.
If you think about it, the wayit worked in the past few years,
we still use different platformsand computers.
(27:16):
And the way it worked is thehuman did something and they
used a computer to assist themin what they were doing.
And what's happening right now.
At a faster and faster pace isit's gonna flip itself around.
The computer will be able to domost of the stuff, and it will
need feedback from a human inorder to adjust itself to do the
job right.
(27:37):
Let's talk a little bit aboutwhat's going on right now with
ChatGPT and where it's goingnext, and even a little bit
about OpenAI, the company thatis behind it.
Because it's really importantfor us to understand beyond the
technical implications and thesocial implications and the
economical implications, theforces that are behind this
thing.
So OpenAI, the company that isbehind Dall-E and ChatGPT and
(27:59):
there's a lot of other companiesin this field, but they just
really made a huge splash in thevery recent months.
Open AI is an organization thatwas founded in San Francisco in
2015 by Sam Altman, ReedHoffman, Jessica Livingston,
Elon Musk, Peter Thiel.
So if you don't know thesenames, Sam Altman was the
president of Y Combinator.
Maybe the most known andsuccessful accelerator in the
(28:21):
world.
Jessica Livingston is from thereas well.
Elon Musk, I probably don't needto introduce to you.
Reid Hoffman is the founder ofLinkedIn.
Peter Thiel is a hugeentrepreneur and all of them are
huge investors in the techworld.
And they got together and said,this AI thing is happening.
It's happening fast.
It could do a huge damage in theworld unless somebody does it
the right way and try to protectus humans from it.
(28:44):
And they pledged to togethercreate a nonprofit organization
that will develop, what theycalled a friendly ai.
and they jointly invested 1billion in this initiative.
So that was the way it started.
It started as a way to create anAI that will serve everybody
hopefully while minimizing therisks of this platform.
(29:05):
Somewhere along the way that wassomewhat lost because now
there's a for-profitorganization on the side of the
nonprofit organization thatfor-profit organization got a
really, really big investmentfrom Microsoft of about a
billion dollars to get it towhere it is today.
And they've already committed tospending what speculations say
are additional$10 billion ofinvestment, which is obviously
(29:27):
an insane amount of money.
So you're asking yourself, whyis Microsoft investing in a chat
platform?
And the reason is that's theirway to fight Google.
They're planning on integratingChatGPT into more or less
everything, Microsoft.
So the three biggest layers ofthat.
One is obviously Bing, which isa search engine that has been
(29:47):
getting, its ass kicked byGoogle because it cannot
compete.
But now with this thingintegrated into it, they could
provide a whole different kindof experience in their search
engine that may be appealing tosome people, which has put
Google on a very high alertbecause Google have a similar
platform that they haven'treleased to the public, because
(30:08):
if they do, it's shootingthemselves in the foot when it
comes to the business model ofhow you monetize this versus ads
when you scroll through a feedof search results.
So Google is in a veryinteresting position right now
with how they're gonna respondto what Microsoft is doing.
The other two layers of theMicrosoft stuff that they're
(30:28):
gonna use this for one.
is Azure Azure is theircloud-based hosting platform
that will be able to become alot smarter and provide more
services in a much faster andbetter way by leveraging that.
And the third one is obviouslyMicrosoft Office.
Think about Microsoft Word.
Instead of being a wordprocessor, it become a word
generator, which is obviously amuch more effective and
(30:50):
impactful tool to anybody who'susing it.
And the same thing withMicrosoft Excel, because you'll
be able to tell it what you'retrying to create without really
knowing how to write all theequations in the background.
And it will build it for youvery quickly, very effectively,
exactly tailored to your needs.
and that's why Microsoft isinvesting this huge in insane
amount of money into ChatGPTChatGPT is based on the GPT
(31:14):
engine that's called GPT-3.5which is part of the third
release of the GPT engine byOpenAI, and they're planning to
release GPT-4 in the nearfuture.
Everybody's expecting it tostill be in Q1, this is recorded
sometime in January.
So we're talking about the nexttwo to three months.
What's the difference?
The biggest difference is theamount of data that they've used
(31:35):
in order to train the model.
The previous model, meaning theGPT that is used today has 175
billion parameters it's trainedwith, and GPT-4 is going to be
trained with a 100 trillion.
So more or less, three orders ofmagnitude, more data has been
put into it to make it evenbetter, even more capable, even
(31:58):
more agile than ChatGPT istoday.
And again, that's happening inthe next two to three months.
If what you heard so far,whether you knew some of it or
not, is make your head hurt andis making your head spin.
That was the intention of thisepisode.
Really.
I'm not trying to give all theanswers.
I'm trying to just describe thecurrent situation.
(32:19):
I'm also saying that I haveobviously no clue and I don't
think anybody does.
What are the social oreconomical implications?
of all of this.
What I do know is that the trainhas already left the station.
There is no coming back, andwe've passed the point of no
return.
So what does that mean to you asa person and to the business
you're in.
(32:40):
It means that you need to beaware that this is happening.
You cannot play the ostrich andput your head in the sand
because this is happening.
And it's happening at a fasterpace than anything else in
history, and it's acceleratingbecause the capabilities
themselves allow to generate thenext variation faster.
So what can you do about it?
(33:01):
Research, use these tools everysingle day.
Find use cases like the onesthat I've described, or just
look online for other use casesand start playing with these
tools.
Most of them today are eitherfree or very, very close to
free.
There's like a fractional smallfee that you have to pay to use
it at a larger scale.
They are APIs so you canintegrate it into things in your
(33:21):
ecosystem that costs a littlebit of money.
What will give you an incrediblecapability.
So all I'm suggesting is beaware, stay connected, play with
the tools and be ahead of thepeople who do not do that
because this tsunami is comingand you wanna be on probably the
best raft there is when it hitswhere you are versus be
(33:42):
completely caught off guard.
That's it for today.
This is not only bad, likeanything else that happened in
history that presented new toolsand new capabilities to
humanity.
Humanity adapted and ended upbeing better than it was before.
Once it's learned to use thetools, so this may sound all
doom and gloom, it's actuallythe next evolution of the human
(34:03):
race.
It's just happening faster thananything we've seen before.
And hence why I'm suggesting beinvolved, be active, learn the
tools, you'll come up ahead Andif enough people do that, we
will make the world a betterplace than it is right now.
Leveraging tools that will makeour life easier and more
impactful and more effective anduntil next time, have an amazing
(34:25):
week.
Hello and welcome to the Business Growth
Accelerator.
This is Isar Meitis, your host,and this is a very special
episode.
I've been working on theresearch for this for a while,
and today we're going to talkabout AI, artificial
intelligence, and how it'simpacting and how I think it's
going to impact, the future ofthe world and more specifically
the future of business.
(01:25):
And what's been happening in thepast few years is that AI has
been getting into more and moreaspects of our lives and more
and more aspects of ourbusiness.
And 2022 has been the tippingpoint where things started
exploding, but before we jumpinto 2022, let's figure out what
happened before and talk alittle bit about concepts.
(01:46):
Most of these episodes is gonnabe very practical with use cases
that people are using AI thatyou can use in your business.
But some background I think willbe helpful, especially to people
who has not been reading orfollowing this world.
Artificial intelligence Is apart of computer science that
allow computers machines tolearn in a similar way that
(02:08):
humans learn.
Meaning instead of writing apiece of code that tells the
computer to do something, itactually learns from mistakes or
from feedback that it's beinggiven based on action that it's
taking.
So there's a whole process oftraining the machine for it to
learn, just like you would trainkids or adults in order to learn
(02:28):
new things.
You give it feedback and then itknows what was wrong and what's
better, and you keep on doingthis again and again and again,
and that's how the machinelearns.
Hence, machine learning is theconcepts behind artificial
intelligence.
When we say artificialintelligence, it's a really
broad term and there's many moregranular terms of this that has
(02:49):
to do with different aspects ofwhat it's actually going to do.
Meaning is it's gonna deal withlanguage processing, is it going
to deal with image processing,video processing, data analysis,
and so on.
Each and every one of them is adifferent aspect of artificial
intelligence and machinelearning and people today are
talking about general ai, whichmeans the capability for a
(03:11):
machine to actually be like ahuman, meaning, be able to do
all the things that we do andnot be specialized at one thing.
And there's still a few yearsbefore that will be possible,
but even the capabilities thatexist today are pretty
incredible.
And we'll touch on a lot of themin this episode.
AI has been used even by youmaybe even without knowing about
it for a while.
(03:32):
If you've been shopping onAmazon, which most of us
probably have, and you seerecommendations of other stuff
you might wanna buy, or if yougo to Netflix and you get
recommendations of the nextvideo, or if you go on Google
and you get results.
All of these things are based onartificial intelligence, meaning
it's taking a huge amount ofdata and it's looking at your
personal touchpoints anddifferent data it has on you,
(03:54):
and it's trying to predict basedon that what's gonna be the most
relevant and that will attractyou to take a particular action.
So you've been interacting withAI systems for a very long time.
What happened in the past fewyears is that AI from being kept
only for giants like Microsoftand Netflix, more and more
companies and startups startedreleasing AI driven tools that
(04:18):
allows companies and individualsto use and leverage AI for
different business and personaltasks.
Let's take a few examples.
copy.ai has existed for a while.
It's a platform that allows youto, like the name suggest, write
copy.
So marketing copy, like websitecopy, newsletter copy, social
media, post copy, et cetera.
(04:38):
It allows you to do all thesethings and it'll write the copy
for you based on differentparameters you give to it.
Another example is RYTR.
It is spelled R Y T R and thewebsite is rytr.me and it allows
you to write longer kind ofcopy, like even blog posts and
longer piece of content based onparameters you give it.
(05:00):
If you want to learn more aboutthese kind of tools, I've done a
whole episode dedicated to RYTRand similar platforms.
It's episode 126 of thispodcast, so just go back and
check that one out.
Such tools became more and moreavailable that started doing
more and more unique and coolthings.
The next interesting iterationwas released by a company called
OpenAI, which you're gonna talkabout more later on in the show.
(05:23):
And OpenAI back in January of2021, released a tool called
Dall-E.
Obviously wink, wink to Dali thepainter.
And what it does, it enables youto put in a text, a descriptive
text, and it creates an imagefor you right there on the spot.
So it doesn't find an image onthe internet.
It literally creates it in realtime based on parameters.
And you can use whateverparameters you want.
(05:44):
And when I say parameters, itsounds very techy, but all you
have to do is type what you wantto paint.
So if you'll say, I want you topaint a house on a moon.
It will paint a house on a moon.
But you can be a lot moredescriptive than that and say, I
want you to paint a house on themoon at sunrise in a realistic
way.
It's gonna generate almost likean image, like picture of a
(06:05):
house on a moon.
But you can also say, I want thehouse to be two stories high and
to have yellow windows, and Iwant three cars parked in the
garage, and it will do that aswell.
Or you can go and say, I wantthat whole thing to be painted
in the Renaissance style, and itwill give you that kind of style
to the painting.
So it's really limitless to thetype of data you can give it,
(06:29):
and the more data you can giveit, the more specific it will
become.
And then gives you options andyou can pick one options and
keep on changing that until youget to an outcome that you want.
These kind of tools startedbecoming more and more available
for video processing, for imageprocessing, for written content
creation.
And so on.
(06:49):
These kind of tools existed fora while and more and more of
them started coming out.
But something dramatic changedin November of 2022.
In November of 2022, OpenAIreleased two tools.
One is Dall-E 2, so it's a moreadvanced form of the first
version of this that is just aimage generator based on text,
(07:11):
but the thing that reallychanged everything was ChatGPT.
ChatGPT is based on their chatengine that allows you to chat
with their bot that they've beentraining for a very long time.
And the reason I'm saying itchanged anything is the fact
that you can literally give itany task that has to do with
(07:32):
organizing content, writingcontent, interpreting content,
creating new content in anywritten form you can imagine.
So this could be a blog post,this could be a legal document.
This could be a song.
This could be a poem.
This could be a summary inbullet points.
This could be computer code.
Literally any format you ask itto produce something, it will
(07:52):
produce it in a very highquality.
And in seconds, and at least fornow for free.
And when I say everythingchanged with the launch of
ChatGPT, I just wanna talk alittle bit about adoption.
You probably heard the termChatGPT a million times in the
past two months.
But if we're talking about amillion, let's talk about how
quickly it took differentplatforms to get to a million
users.
(08:13):
Netflix took about 3 1/2 yearsto get to a million active
users.
Spotify took 5 months.
Facebook took 10 months.
So these are the timeframes ofsome of the best, most
successful adoptions of newplatforms in the world that we
live in.
It took ChatGPT 5 days to get toa million active users, and this
(08:35):
has been continuing to grow atthe same crazy pace.
Now there's been chat platformsbefore that.
So what's the big deal aboutChatGPT and why is it taking the
world by a storm?
The first thing is it's a verysimple and easy to use user
interface.
They're using the same conceptthat Google used when they
(08:55):
launched their search, it's justone line.
There's no menus, there's nobuttons, there's nothing you can
do other than write someinstructions in the line and get
a response and chat back andforth with the chat bot.
So it's extremely easy to use.
But the other thing is that it'sreally extremely powerful and
very, very versatile.
It will do really almosteverything you ask it to do on a
(09:18):
very wide range of topics andtasks that you will ask it to
perform for me, like I mentionedbefore.
So what can you do with it?
First of all, you can writecopy.
So if you need a new concept foryour website, you can ask
ChatGPT to write it for you.
Write me three differentvariations of the content I
(09:38):
should put on my website if I amtrying to attract mechanics to
work for my shop.
And it will give you threedifferent options of marketing
content that you can put on yourwebsite.
You can use it to write copy forsocial media posts, you can take
whatever topic you wanna writeabout this week and say, I would
like you to write a post about.
(09:59):
1, 2, 3 and 4 for LinkedIn andyou will create the content to
be adopted in the right lengthand the right formatting to fit
a LinkedIn post.
You can go a step beyond thatand use reference material as
the baseline for the post.
So let's say you wrote a blogpost and you wanna summarize it
to a post on social media.
You can copy and paste thecontent that you've written, the
(10:21):
long form content, paste it intoChatGPT and say, I want a
summary of this as a post onLinkedIn, and it will do it for
you.
You can ask for it as a summaryin bullet points, and it will do
it for you.
And if talking about summaries,it can summarize literally
anything.
If you have a very long documentyou wanna summarize, you can
copy and paste the document intoChatGPT, ask it to summarize for
you in whatever length orwhatever definition you want.
(10:43):
So I want 20 bullet points as asummary of this whole thing, and
it will do it for you in a veryaccurate and concise way.
You can give it a url.
Let's say you want a summary ofa specific blog post.
You can go and copy the url ifthat blog post, drop it into
ChatGPT and ask it to summarizeit for you, and it will.
Now, that sounds a little bitlike cheating, but not really
because we think about back inthe day when we had to read
(11:05):
specific, really long, boringbooks for graduation.
Most people did not really readthe full books.
They read a summary that youcould buy somewhere at a
bookstore that will actuallyallow you to read 60 pages
instead of 300 pages and stillbe able to pass the exam.
So it's kind of the same conceptonly for any kind of content you
want.
The other cool thing about it isthe more information you're
gonna give it, the more it'sgonna use that information in
(11:28):
order to make your content morespecific.
If you use specific keywords, ifyou use specific ideas, if you
mention specific things, it willuse those in the outcome of the
content that it is generating.
So, Creating any kind of writtencontent, whether it's original
from scratch, based on ideas andconcept, whether it's a summary
of something, whether it's withreference to something, becomes
(11:50):
an effort of seconds instead ofsometimes hours or days.
Another thing that he doesreally well is to write pieces
of code, computer code.
So if you're a computerprogrammer or you're learning
how to do that, you can say whatyou want it to create, give it
the language.
So let's say you want a codingin Java to do something, you
will describe that something andyou will create the Java script
(12:12):
for you to use.
In your code in order to do thatspecific thing.
Now can you program a completeplatform?
Not yet.
I'm sure we'll get there, butprogramming or solving specific
problems with code, it knows howto do that extremely well,
extremely efficient, extremelyfast, and very accurate.
Now take it to the day-to-daybusiness world, you can write
(12:32):
the company update with that.
You can say, these are thethings I wanna talk about.
Can you write me, the speech?
You can use it to write otherspeeches that you wanna put on
stages with specific informationthat you feed into it, just to
break it down into the rightformat, the right structure, the
right wording.
Whatever you want help from itto do, it can do for you.
(12:52):
You can use it for marketresearch.
If you want to know somethingabout your industry or a new
industry you want to go into,you can use ChatGPT for that as
well.
You can go in and say, what arethe 10 largest companies in this
and that niche?
And it will give you the list ofthe companies and then you can
go one by one and say who arethe companies that are most
known to use that particularproduct?
(13:14):
What is that product most knownfor?
What kind of marketing languagehas this company used in order
to attract the people that it'sbeen attracting and so on and so
forth?
You can keep on peeling theonion and drilling deeper and
deeper to find more and moreinformation that before was a
lot harder to get to because youhad to run a Google search and
then look at 50 differentarticles and then from that,
(13:36):
find the information.
But instead, This thing is doingit for you and it's doing it in
again, seconds instead of hoursof research that you have to do.
I'll give you one little warningthough.
ChatGPT is not connected to theinternet.
It's an offline platform, andyes, it's been trained based on
a lot of internet data and thenew variation of it, which is
coming out soon.
(13:57):
We'll have a lot more data, butit's not really online.
As an example, it does not knowthat Elon Musk has bought
Twitter, it does not knowthere's a war between Russia and
the Ukraine.
If you're looking for up to dateinformation, you will not get
it.
So it has some disadvantages andlimitations, but on a very big
level of doing research.
It's an incredible, incredibleresearch tool.
(14:17):
And like I said, you can drilldeeper, you can get summaries,
you can do whatever you wantwith just typing the right
questions.
Next thing it does extremelywell is customer support.
There's more and more platformsbeing developed on top of AI
solutions to do customerservice.
Why?
Because customer service is alogical process.
There's something that happened.
There is an X amount ofpotential outcomes.
(14:40):
There's the way to navigatethem, and AI is very, very good
at doing that.
What we'll see in the nearfuture is more and more customer
service processes being takenover by AI versus actual people,
whether it's a live call, chat,whatever the case may be, that
will be taking over, it'salready happening today, right?
So when we call an ivr, whichtoday is really dumb and
(15:00):
annoying and we hate it, but inthe very near future, it will be
just like speaking to a person,but different than today, that
person will know literallyeverything they need to know and
will have access to everypotential outcome, and will
really be able to navigate veryquickly to the potential outcome
based on the set of guidelinesthat it has.
Same concept obviously appliesto sales, so you can take your
(15:21):
company's knowledge on how tosell your product or service
best.
Train an AI model and that useChatGPT as an engine if you
want, behind the scenes.
Then when people use chat onyour website, they can get very
accurate, very sophisticated,very to the point answers to
their questions without actuallyhaving to deploy salespeople to
(15:42):
do these things.
You can also use it for HRrelated stuff in your company,
like planning the next Halloweenparty.
If you go into ChatGPT and say,I have a hundred employees and I
wanna plan a Halloween partythemed for the eighties, what do
you suggest we would do?
And it will give you a list ofeverything you should do, what
drinks you should prepare, whatcostumes people should wear, how
(16:04):
big of a venue you need to rent,and so on and so forth.
It will help you plan the party.
The options are almostlimitless.
Wanna take it up a notch?
Let's look at some other usecases that you can use in your
business in order to do somemore interesting things.
Let's assume for a second youhave a product or a service and
you want to create demo videosof how to use that particular
(16:27):
thing or specific aspect of itand so on.
The way to do this today, if youwant a person involved is to
rent a studio or build a studioin your office and have cameras
and lighting and spend a lot oftime and a lot of money
recording people in front ofcameras that make mistakes and
have to go back and recordagain, and they get tired and
they get confused.
And then you have to get allthat content and edit it so it's
(16:49):
perfect and it looks great, andthen you can use it in whatever
how-to videos that you want.
Well, that is gonna disappearvery, very quickly because today
there are already tools outthere like Synthesia and D-ID
and a bunch of others thatreally enable you to create a
very realistic avatar.
(17:10):
And when I say very realistic,it looks like people, it talks
like people.
You can still tell with some ofthem it's not human.
It's still good enoughdefinitely for any demo of the
product.
And it takes seconds because allyou have to do is type the text
into it choose an avatar to yourliking.
It doesn't look like a cartoon.
It's an actual person.
It looks like a real person.
(17:31):
And this person, because it's asynthetic person, can be in any
age, any ethnicity you choose,any kind of features that you
would like.
You really have endless kind ofoptions and it can speak almost
any language on the planet.
So you copy and paste.
Text into there and it will sayit perfectly without mumbling,
(17:53):
without going, uh, uh, uh, inthe middle.
So you have to cut and recordagain without spending studio
time, without spending editingtime, and you have a perfect
video in seconds.
Now, let's combine these twothings together.
Let's say you want to create apresentation about a topic.
In video to put it on yourwebsite, to put it on social
media, to put it wherever youwanna put it.
(18:15):
Instead of working in front of acamera, instead of recording all
the content, instead of editingall the content, instead of
doing the research for thecontent, you can go to ChatGPT
and literally give it the inputand saying, I want to create a
lecture on these topics withthis information in mind.
Paste in the information youwanted, and the lecture should
be 45 minutes and I need it inEnglish or Spanish or Italian,
(18:38):
and you will generate it foryou.
And then you take thatinformation to one of the
platforms that I mentionedbefore, like Synthesia or D-ID
or one of those Pasteinformation in there.
Choose your avatar and you havethe video ready to go.
A whole game changer in the waycontent is being created.
(18:58):
So far we talked about creatingwritten content, and we talked
about creating video content.
Now let's talk a little bitabout images and how, again,
these things can combinetogether.
As I mentioned earlier, whenChatGPT was released, Dall-E 2
was released, and as thepopularity of these tools
skyrocketed, other platformsthat are AI image generators
(19:19):
skyrocketed with it.
The two biggest ones togetherwith Dall-E are midjourney and
Stable Diffusion.
All of these basically do thesame thing as I mentioned
before, you give it adescription as detailed as you
want of what you want created,and you will create an image for
you.
That's obviously a very cooltoy, but it's.
Also a very impactful andeffective business tool.
(19:41):
Why?
First of all, because right nowmost companies are paying a huge
amount of money to differentstock photos websites in order
to get specific photos that theywant to use in their marketing
on their website, because theyhave to be licensed for the
people who took the pictures.
But now you can create whateverimage you want in whatever
scenario, in whatever lighting,in whatever concept you want on
(20:01):
the fly, which means you don'treally need stock photo, you
don't need to search for stockphoto.
You can actually create theactual images that you want as
needed per the exact scenariothat you want to use them for.
That's one way to use it.
Another way to use it is indocumentation that you're trying
to create.
If you wanna talk about specificpeople in specific scenarios and
you want to use images thatdescribe that, you can do that
(20:25):
and create those images rightthere and then, instead of going
and searching through hundredsof photos in order to find the
one that you want.
Another interesting and coolimplementation that people
already started doing is peoplestarted creating kids books in
minutes.
And how do they do that?
So you go to ChatGPT and youwrite I wanna write a kid's book
(20:46):
about two brothers and a sister,and you can give them your
specific names that are learninghow to be green and how to
recycle and how to make theworld a better place and stop
global warming.
And I want it to be 10 pageslong, and it will create the
text for you in rhymes accordingto all the parameters you said.
And now for each of those pages,for each section of rhymes about
(21:08):
a specific topic, you go to oneof those image generators and
you describe what the text says,and it will create images for
you.
And you can say, these are forkids' books, and they will be
book friendly kind of drawings.
Now all you have to do is open aPDF editor, place a page from
the written content in front ofit on the other side, place the
(21:29):
image that you just created.
Save the pdf, upload it toAmazon, and you have a kid's
book on a specific topic thattook you minutes to generate.
Now, this may not be your thing,but it gives you an idea of what
use cases you can use this forand how extremely efficient the
process becomes.
The next aspect of images isobviously graphic design.
(21:51):
People use graphic design formultiple different things.
From designing websites, todesigning logos, To designing
brochures and so on and soforth, designing products, and
so far the people who did thiswere graphic designers who got
trained and got experienced andspent years doing so.
Today almost every person can dothese things.
To be fair, most of them are notgreat in graphic design.
(22:14):
Based on most of everything thatI'm reading right now and
researching right now, design isprobably the first complete
revolution of the world whereliterally anybody will be able
to create incredible designswith a few sentences, and then
changing it around until theylike them.
It will be a much faster, muchmore efficient, and a way
cheaper process than it is todaywith graphic designers.
(22:35):
The predictions are that thiswill be really good and really
accessible this year, meaningstill in 2023.
So let's do a quick recap.
This whole concept that I talkedabout so far, it's called
generative ai.
Generative AI allows you to likeits name, so just generate new
things using ai.
This could be video, this couldbe audio, this could be written
content.
And Sequoia recently released aresearch that they've done and
(22:59):
their predictions based on thatresearch says that most likely
100% of the content that we'llconsume by the year of 2030 will
be created by AI versus actualpeople, roughly at the same
time, potentially before that,the research says all computer
code will be AI driven and notwritten by people.
(23:20):
As I mentioned, the sameresearch says that by the end of
2023, graphic designcapabilities will be strong
enough.
It still won't replaceeverything, but it will be
capable enough to replace everygraphic design need.
Now, so far, I talked aboutstuff that is very, very
specific.
It's content creation ofdifferent kinds, right?
So people who are writers,people are technical writers,
(23:41):
people are graphic designersand, and so video producers and
so on.
So you may think, okay, whatmight be protected from this
change?
Probably stuff that is veryhuman, high touch, emotional
kind of thing, I got anothersurprise for you.
A guy named Dr.
Robert Morris, who has a PhDfrom MIT.
(24:02):
He's the founder of Koko.
Koko is an an online emotionalsupport platform.
So it's a chat platform thatenables people, mostly young
adults that have emotionalissues.
To have a chat with somebody whowill support them through their
difficulties, it's an incredibleplatform.
He ran a test with over 4,000people who approached his online
(24:23):
service with more than 30,000messages.
And the way he ran the research,he had a person, professional
human therapist, the people healways uses, but instead of
having them respond to theparticipants, he copied the
participant's question orsentence or whatever they wanted
into ChatGPT and copied andpaste it back the respond to his
(24:47):
participants, who, again, werelooking for emotional support
through chat.
The chat messages get scored,and at the end of each chat
people say how good they feelabout the responses they got and
how much it helped them.
And the ChatGPT messages scoredsignificantly higher than the
human therapists.
In addition to that, the respondtime dropped by 50% and that's
(25:09):
while there was a human in theloop, just to making sure that
Chat PGT doesn't send backsomething that is actually bad
or inappropriate or couldactually harm the person on the
other end.
So if you think about runningthis at scale, without the human
in the loop, you can get toextremely fast respond time.
That is still extremely helpfulthat people that are looking for
(25:30):
support find more helpful thanprofessional human therapists.
It's hard for me to think of ascenario that is requiring more
emotional empathy than thisparticular scenario.
So even this is not protected.
By the way, once hisparticipants learned that
there's a machine in the loop,which they asked them if they
(25:52):
would use it, they would say, ifit was a machine, we would not
use the platform, even thoughthat the vast majority said that
the responses they got wereactually better than the ones
that the human gave them.
Again, they did not know whatwas human and what was Chat PGT
producing.
But it's just coming to show youthe power of this platform.
This may sound like the end ofthe world, right?
More or less, every professionwe know is gonna go away because
(26:13):
the robots are going to replaceit.
The reality is very different.
And first of all, let's go backand look at history and probably
the closest thing to this is theindustrial revolution.
But it's a different scenariothan the Industrial Revolution
because the IndustrialRevolution took years and years
to happen and take over theworld, and this will take
between months and a couple ofyears to do a lot of things
(26:35):
we're doing today.
So the speed in which it'shappening is way faster than the
Industrial Revolution.
The other big difference isbecause so many platforms are
developed on top of AI solutionstoday, while the Industrial
Revolution touched on specificprofessions, this is gonna touch
more or less anyone on theplanet, regardless of what
(26:55):
you're doing today.
some professions faster thanother, obviously, as I mentioned
earlier.
But as I mentioned, theseplatforms have to be trained,
meaning there has to be somebodytraining those platforms,
helping them get better andbetter and better so they can do
all these tasks.
If you think about it, the wayit worked in the past few years,
we still use different platformsand computers.
(27:16):
And the way it worked is thehuman did something and they
used a computer to assist themin what they were doing.
And what's happening right now.
At a faster and faster pace isit's gonna flip itself around.
The computer will be able to domost of the stuff, and it will
need feedback from a human inorder to adjust itself to do the
job right.
(27:37):
Let's talk a little bit aboutwhat's going on right now with
ChatGPT and where it's goingnext, and even a little bit
about OpenAI, the company thatis behind it.
Because it's really importantfor us to understand beyond the
technical implications and thesocial implications and the
economical implications, theforces that are behind this
thing.
So OpenAI, the company that isbehind Dall-E and ChatGPT and
(27:59):
there's a lot of other companiesin this field, but they just
really made a huge splash in thevery recent months.
Open AI is an organization thatwas founded in San Francisco in
2015 by Sam Altman, ReedHoffman, Jessica Livingston,
Elon Musk, Peter Thiel.
So if you don't know thesenames, Sam Altman was the
president of Y Combinator.
Maybe the most known andsuccessful accelerator in the
(28:21):
world.
Jessica Livingston is from thereas well.
Elon Musk, I probably don't needto introduce to you.
Reid Hoffman is the founder ofLinkedIn.
Peter Thiel is a hugeentrepreneur and all of them are
huge investors in the techworld.
And they got together and said,this AI thing is happening.
It's happening fast.
It could do a huge damage in theworld unless somebody does it
the right way and try to protectus humans from it.
(28:44):
And they pledged to togethercreate a nonprofit organization
that will develop, what theycalled a friendly ai.
and they jointly invested 1billion in this initiative.
So that was the way it started.
It started as a way to create anAI that will serve everybody
hopefully while minimizing therisks of this platform.
(29:05):
Somewhere along the way that wassomewhat lost because now
there's a for-profitorganization on the side of the
nonprofit organization thatfor-profit organization got a
really, really big investmentfrom Microsoft of about a
billion dollars to get it towhere it is today.
And they've already committed tospending what speculations say
are additional$10 billion ofinvestment, which is obviously
(29:27):
an insane amount of money.
So you're asking yourself, whyis Microsoft investing in a chat
platform?
And the reason is that's theirway to fight Google.
They're planning on integratingChatGPT into more or less
everything, Microsoft.
So the three biggest layers ofthat.
One is obviously Bing, which isa search engine that has been
(29:47):
getting, its ass kicked byGoogle because it cannot
compete.
But now with this thingintegrated into it, they could
provide a whole different kindof experience in their search
engine that may be appealing tosome people, which has put
Google on a very high alertbecause Google have a similar
platform that they haven'treleased to the public, because
(30:08):
if they do, it's shootingthemselves in the foot when it
comes to the business model ofhow you monetize this versus ads
when you scroll through a feedof search results.
So Google is in a veryinteresting position right now
with how they're gonna respondto what Microsoft is doing.
The other two layers of theMicrosoft stuff that they're
(30:28):
gonna use this for one.
is Azure Azure is theircloud-based hosting platform
that will be able to become alot smarter and provide more
services in a much faster andbetter way by leveraging that.
And the third one is obviouslyMicrosoft Office.
Think about Microsoft Word.
Instead of being a wordprocessor, it become a word
generator, which is obviously amuch more effective and
(30:50):
impactful tool to anybody who'susing it.
And the same thing withMicrosoft Excel, because you'll
be able to tell it what you'retrying to create without really
knowing how to write all theequations in the background.
And it will build it for youvery quickly, very effectively,
exactly tailored to your needs.
and that's why Microsoft isinvesting this huge in insane
amount of money into ChatGPTChatGPT is based on the GPT
(31:14):
engine that's called GPT-3.5which is part of the third
release of the GPT engine byOpenAI, and they're planning to
release GPT-4 in the nearfuture.
Everybody's expecting it tostill be in Q1, this is recorded
sometime in January.
So we're talking about the nexttwo to three months.
What's the difference?
The biggest difference is theamount of data that they've used
(31:35):
in order to train the model.
The previous model, meaning theGPT that is used today has 175
billion parameters it's trainedwith, and GPT-4 is going to be
trained with a 100 trillion.
So more or less, three orders ofmagnitude, more data has been
put into it to make it evenbetter, even more capable, even
(31:58):
more agile than ChatGPT istoday.
And again, that's happening inthe next two to three months.
If what you heard so far,whether you knew some of it or
not, is make your head hurt andis making your head spin.
That was the intention of thisepisode.
Really.
I'm not trying to give all theanswers.
I'm trying to just describe thecurrent situation.
(32:19):
I'm also saying that I haveobviously no clue and I don't
think anybody does.
What are the social oreconomical implications?
of all of this.
What I do know is that the trainhas already left the station.
There is no coming back, andwe've passed the point of no
return.
So what does that mean to you asa person and to the business
you're in.
(32:40):
It means that you need to beaware that this is happening.
You cannot play the ostrich andput your head in the sand
because this is happening.
And it's happening at a fasterpace than anything else in
history, and it's acceleratingbecause the capabilities
themselves allow to generate thenext variation faster.
So what can you do about it?
(33:01):
Research, use these tools everysingle day.
Find use cases like the onesthat I've described, or just
look online for other use casesand start playing with these
tools.
Most of them today are eitherfree or very, very close to
free.
There's like a fractional smallfee that you have to pay to use
it at a larger scale.
They are APIs so you canintegrate it into things in your
(33:21):
ecosystem that costs a littlebit of money.
What will give you an incrediblecapability.
So all I'm suggesting is beaware, stay connected, play with
the tools and be ahead of thepeople who do not do that
because this tsunami is comingand you wanna be on probably the
best raft there is when it hitswhere you are versus be
(33:42):
completely caught off guard.
That's it for today.
This is not only bad, likeanything else that happened in
history that presented new toolsand new capabilities to
humanity.
Humanity adapted and ended upbeing better than it was before.
Once it's learned to use thetools, so this may sound all
doom and gloom, it's actuallythe next evolution of the human
(34:03):
race.
It's just happening faster thananything we've seen before.
And hence why I'm suggesting beinvolved, be active, learn the
tools, you'll come up ahead Andif enough people do that, we
will make the world a betterplace than it is right now.
Leveraging tools that will makeour life easier and more
impactful and more effective anduntil next time, have an amazing
(34:25):
week.
Popular Podcasts
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.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.