July 31, 2025 • 43 mins
In this episode of Imperfect Marketing, host Kendra Corman sits down with Nuri Cankaya, VP of Commercial and AI Marketing at Intel and author of AI in Marketing, to break down how artificial intelligence is reshaping marketing, productivity, and the very fabric of modern business.
With over two decades in the AI space—including 17 years at Microsoft—Nuri shares how today’s marketers, business owners, and professionals can leverage AI to work smarter, serve customers better, and prepare for a future driven by agentic AI and beyond.
Through practical examples and bold predictions, he offers a roadmap for how to ride the AI wave—ethically, securely, and strategically.
We Explore:
The Role of AI in Modern Marketing
- Why AI isn’t new—and why ChatGPT is just a moment, not the movement
- How AI streamlines campaign creation, personalization, and data-driven decision making
- Examples of multimodal AI tools changing how marketers ideate, design, and launch faster
The Co-Creation Principle: AI Then Eyes
- Why human oversight is critical for AI-generated content
- How to treat AI like an “intern with unlimited hours”—great at speed, but still needs supervision
- Why prompt engineering and context are the secret ingredients to meaningful outputs
Enterprise-Ready AI: Security, Customization & Scale
- Ways big organizations are building private AI instances with brand-specific training
- The importance of developing generative AI usage guidelines for teams
- How hybrid models can provide power without compromising IP or data security
The Rise of Agentic AI, AGI & Ethical Boundaries
- What agentic AI and AGI (Artificial General Intelligence) really mean—and why they’re closer than we think
- Three major ethical risks to watch: plagiarism/IP, black-box decision-making, and corporate responsibility
- Why traceability and transparency in AI outputs matter more than ever before
🔑 Key Takeaways for Marketers and Business Leaders
- AI won’t take your job—but someone who uses it better might.
- Sell outcomes, not features. Customers want solutions, not specs.
- Secure, contextual AI use is the next competitive advantage.
- Upskilling is non-negotiable. The next wave of AI will reward those who learn, test, and adapt.
Whether you're a solo marketer, a tech leader, or a curious business owner…
…this episode offers clarity, inspiration, and real-world advice on how to integrate AI into your workflow without losing the human touch that makes great marketing work.
🎧 Ready to ride the AI wave instead of being crushed by it?
Tune in to learn how to blend strategy, ethics, and innovation in an AI-powered world.
📚 Connect with Nuri Cankaya & Grab the Book
- LinkedIn: https://www.linkedin.com/in/nuricankaya/
- Follow his daily insights on LinkedIn for AI + marketing inspiration
- Book: AI in Marketing on Amazon
Looking to leverage AI? Want better results? Want to think about what you want to leverage?
Check and see how I am using it for FREE on YouTube.
From "Holy cow, it can do that?" to "Wait, how does this work again?" – I've got all your AI curiosities covered. It's the perfect after-podcast snack for your tech-hungry brain.
Watch here
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Hi, I'm Kendra Korman .
If you're a coach, consultantor marketer, you know marketing
is far from a perfect science,and that's why this show is
called Imperfect Marketing.
Join me and my guests as weexplore how to grow your
business with marketing tips and, of course, lessons learned
along the way.
Hello, and welcome back toanother episode of Imperfect
(00:27):
Marketing.
I'm your host, Kendra Korman,and I am super excited to be
talking about one of my favoritetopics.
Of course, that is AI, and myguest today is Nuri.
He is the VP of Commercial andAI Marketing at Intel Corp and
recently wrote a book about AI.
But we are going to be talkingall about AI today.
(00:47):
Thank you so much for joiningme today.
Why don't you tell us a littlebit on how you got into AI?
Speaker 2 (00:53):
Absolutely.
Thanks a lot for having me,kendra, on the show.
So I was working on AI when Iwas in college, like almost 25
years ago, but, like AI changeda lot, especially in the 2018-20
timescale.
We have seen a lot of progresson large language models.
Many people, I think, willrefer to ChatGPT as a movement.
(01:15):
I was working at Microsoft for17 years and I worked on data
and AI marketing and that was myteam, working closely with
OpenAI team, and really havingthat moment really inspired me
to work around AI and like howthis is going to change almost
everything, but specifically thearea of marketing, which I'm
(01:38):
working on.
So, yeah, like I mean, I'm withIntel for the last two years
really infusing AI into everywork that we do on the marketing
, but also working with ourpartners and making sure
everybody benefits from the AIand really accelerates the
business outcomes of, like youknow, waiting faster, saving
more dollars and, of course,executing really timely and fast
(02:02):
.
Speaker 1 (02:02):
Okay, so I love that and I have to like just comment
so you have been working on AIfor a long time, because people
ask me all the time whenever I'mpresenting about AI or talking
about it, they're like, well,it's only been around a couple
of years and I was like, youknow, like I've met people who
majored in it, you know, backwhen they were in college, and
some of it was college before me.
You were in college while I wasthere, but that's so cool that
(02:25):
you were working on it for solong.
And really, yeah, I mean peoplelook at the unveiling of
ChatGPT as like this moment intime where AI like happened and
it's like, no, it didn't Right,it's been around for a really
long time and actually many ofthe use cases we see today, not
(02:46):
all of them, are generative AI.
Speaker 2 (02:48):
So, again, like gen AI is a term that we get used to
, but the basics of AI, which islike starting with machine
learning and it's all about thealgorithms learning from each
other.
So, and then many of thetechniques that we use today in
marketing, they are actually thebasics of AI from like 2010s,
2012.
So, again, like I mean, I knowI mean we will speak a lot about
(03:11):
Gen AI and agentic AI, butagain, majority of the things
like on forecasting, predictiveanalytics, customizations, on
campaigns, many of them are justusing the basics.
The better term is narrow AI,but again, it is still AI.
Speaker 1 (03:28):
Yeah Well, and sometimes people are talking
about things and I'm like Idon't even think that's AI, I
think it's that's justprogramming.
Ai has become everywhere, right.
So I couldn't believe this theother day because, you know,
back, I would say, a year or twoago, people were hesitant to
learn about AI and use AIbecause they're like, oh my gosh
, it's going to take my job.
And I was actually doing apresentation a couple of weeks
(03:51):
ago and someone came up to meafterwards and said thank you
for this.
I was hesitant to use AIbecause I was scared it was
going to take my job.
She's like but I really believethat people that know how to
use AI will take my job beforeAI takes my job.
And I agreed with her what areyou seeing on jobs emerging or
(04:11):
changing due to how AI isrolling out?
Because it's rolling out at aridiculous pace, exactly.
Speaker 2 (04:18):
I think your example is a really good one and
especially many of the listenerstoday will have that question
in their mind like, am I gonnalose my job to ai?
I'm an optimist, but again, I Iwant to keep the balance.
So ai is gonna definitelyaffect everybody's life in the
next 10 years period.
Like there's no way that, like,my job is safe, my work is done
(04:41):
.
Like that's why everybodyshould spend some time on
upscaling themselves on AIBecause, as you said, like I
mean, somebody else who's usingAI might take the job.
Also, I feel AI is more of aco-creator for marketing teams,
so which means that, like I mean, it will help you to execute
(05:01):
some of the tasks way faster,way effective, compared to today
or maybe like before.
So that's why I think it'sinevitable for everyone to not
to use AI.
So and I will give maybe acouple of examples but like I
have like a creative team right.
So, like every time we docampaigns, we were like we have
(05:22):
the brand guidelines, we haveall this, from fonts to colors,
to everything that likemessaging goes through.
Now we are using some AI toolswhich we feed the context on.
Like hey, these are all theprevious brand campaigns, the
visuals.
We use the messaging narrativewe developed and it learns from
(05:44):
them.
Use the messaging narrative wedeveloped and it learns from
them.
And then I say, like, usingthis as the baseline, can you
help me to generate the next one?
So, again, maybe just unpackingthat for the listeners.
Ai has like two specific usecases.
One is the training, the otherone is inferencing.
There's a stage in betweenwhich is called the fine tuning.
(06:07):
Just in simple terms, when youtrain AI, it learns from the
experiences and especially withthe latest tools we have, from
like ChatGPT, from OpenAI.
Anthropic has this cloud,google has Gemini, you name it.
Every company has now leadingfrontier models, you name it
like.
Every company has now leadingfrontier models.
You can use them with theirknowledge and you can train them
(06:29):
on your own data, which givesyou the ability to fine tune it,
which means like, hey, intel's,let's say, color is this blue?
And we have the shades of bluewhich are brand appropriate and
everything that you generate itshould be within those
guidelines period, and I tell itvia text.
So I think that's the beauty oflike you're almost talking to
(06:51):
an agency, but it's technicallyAI at the back end and one of
the key things I mean.
I call inferencing.
Really pulling out the data isnot just text anymore, so you
can speak to AI models.
It's called the multimodality.
You can give a voice prompt,you can give like input as a
video and also you can get theoutput as text images, but also
(07:17):
audio files and video files.
And again, like recently, thissummer I think people will
experience a lot of AI generatedvideos.
You cannot differentiate thequality of the video from a real
one versus an AI generated one.
So again, there are a lot ofdebates going on, like is this
the end of the creative rights?
(07:37):
Like who owns the created video?
Is it the AI model or is it thecompany who generated the
content?
But I'm just looking at fromlike it is accelerating our
journey on marketing way better,like just today, for example,
if I go with a traditionalmethod, I brief an agency, I
(07:58):
will wait for them to respond tome two weeks, and then, like
there's a creative person thatneeds to be like getting the
messaging and everything right.
It takes like six to eightweeks from an ideation to
production.
With AI it can be minutes, sothat's the speed agility that we
(08:18):
are talking about.
And then, like I'm reallyexcited about like you can try
different things.
Like, if I do a campaign today,maybe I will have like two
visuals, that's it.
If I know my audience let's sayI have 500 different
personalities, from, like,technical decision makers to
developers I can generate 500different campaign visuals
(08:43):
targeting those 500 differentpersonas.
This wasn't possible before.
So back to your question is AIgoing to take over the jobs?
I don't think so.
If you are using AI properly,you will increase your value in
the market and you will respondto the things way faster than
anybody else.
Speaker 1 (09:02):
I think that that is so important.
It's not taking your job, butthe people that know how to use
it and use it the right way arebecause they can do so much more
and be so much more effectiveand efficient.
One person had told me thatthey had heard the I think it
was the CEO or president ofGoogle say that this will be
(09:24):
more impactful to the world andto people and to humanity than
fire was, just simply because ofhow powerful it is, and I think
that that's just mind boggling.
But when I see it in action andI start to see what's possible,
I start to see what I cancreate as a non-techie person,
(09:45):
right, and how much time itsaves me.
I mean, it saves me 30 to 40hours a week.
That's insane.
Right, I doubled myself.
Right, I don't work any less.
People ask me that all the timeI still work the same amount.
I just get more done, which isfantastic, and no jobs were lost
.
You know things like that.
Speaker 2 (10:02):
This is across the industries.
By the way, I want to underlinethat one because it is not just
marketing.
That is like very visual, likeyou can create, generate things,
but like I see a lot of usagein like healthcare, education.
And those are the maindisruptions, because I believe
(10:22):
the key thing that I'm lookingat is intelligence is becoming
more accessible to everyone.
So, rather than being afraid ofAI taking over the world, now
we are making AI really helpingeveryone on the planet.
So I will give maybe some basicexamples.
But now Google has, andMicrosoft has this medical grade
(10:46):
AI agents and for the listenersjust unpacking that, ai agents
are the agency part of AI, wherethey can work on an like an
agent.
Like they, you can outsourcethings and they work in parallel
.
They find some results and comeback to you when they found
something, or they can workautonomously.
(11:07):
So, as a result of this, likeif you're living anywhere on the
planet, you have access to allmedical results that happened
before.
And if you have an x-ray, forexample, hey, whatever GPT you
are using, can you give me theanalysis as a medical
professional?
And it will give you a detailedanalysis where all the world's
(11:31):
medical doctors combined cangive you the same answer.
So, again, that's pretty likedemocratizing the intelligence
across the globe and this isfascinating.
So you don't have to go to thebest hospital.
You don't have to go to thebest hospital, you don't have to
go to the best university.
Like technically, all thisinformation is accessible to
everyone through this AI tools.
(11:51):
Again, the question is you haveto know how to use them.
So, and not just like basics ofprompt, but like you have to
really master it about, likewhat are the areas that I can
get the benefit of AI in mydaily job?
And, of course, for marketers,going back to the marketing
conversation, you have todocument, like what are you
(12:13):
really doing today?
And like make that assessmentand after that, how can you
automate majority of the thingsby using the right AI tools?
Speaker 1 (12:22):
You have to understand what's capable.
You don't necessarily have tobe able to do it all, right, I
mean, you can bring in otherpeople, but you have to be able
to see those opportunities inyour daily life.
And I liked how you brought insome other things, because some
companies are a little scared ofAI, right, and so they've shut
it down and you can't use it andall these other fun things.
(12:42):
But they're starting to openthat up a little bit more and
more.
They're building their owninstance of ChatGPT right, was
it just their data?
So it's all private.
Yeah, use it in your personallife.
I mean, my sister had a medicalscare last year, last summer,
and I mean ChatGPT was my friend.
I'd take a picture of herresults, I'd upload it in and
(13:03):
I'd say, explain this for aneight-year-old, because I
couldn't read the radiologist'sreport, right, so what does this
mean?
Speaker 2 (13:10):
And it would tell me exactly what that meant and I'm
like, okay, yeah, there's justso much power in it, and so if
you're not able to use it inwork, use it in your day-to-day
life, because there's just againso much power.
That Because there's just againso much power, that's a really
good point, especially for theenterprises.
I'm seeing this more and more,especially with our customers.
That's why there are differentways to use AI in a confidential
(13:36):
and secure way.
So we had it before on thecloud computing as well.
It's called the hybrid AI.
So like going back time, likeagain 2010, this was the year
where there were a lot of withiPads.
Tablets were coming into theworkplace and they were trying
(13:57):
to connect to the work deviceson like netbooks, laptops with
the touchscreen and IT blockedthem.
And then then we call this thebringing your own device moment.
The same happened with theiPhones and Android devices
around 2015.
People love the apps and theybring it, and then you have to
control it.
So same thing is happening withAI today.
People are using ChatGPT,gemini, cloud, perplexity, and
(14:22):
they want to use the samebenefits at work.
So there are ways to make itreally secure and compliant.
The first thing for thelisteners I recommend is setting
up the generative AI guidelinesfor the company.
So this is a job a little biton the.
I hope there will be jobscalled the AI security officers,
(14:42):
so this chief AI securityofficers will define some of
those implementation phases andsome of the models.
Like, january 2025 was a bigmoment for our industry because,
from China, a model calledDeepSeek dropped into the news
and they made it open source,and I think that's the future of
(15:04):
AI.
For many companies, openness isso critical to deploy some of
those solutions and whensomething is open source, you
don't need dependency on cloudsolutions that much, so you can
build the solutions in-house,especially for marketing.
I mean, all the IP that youbuild over years are within your
(15:25):
company firewall and you needthose models not to train the
rest of the world, but your nextmarketing project.
So that's why I think deployingthose solutions locally is as
important as, like, using someof the solutions.
My recommendation would be likejust look out for
implementation options after theassessment stage on like do you
(15:48):
want to keep it completelyoffline for your use?
Do you want to use a hybridmodel where models can get the
latest updates from the cloudbut still keep your data,
especially for training?
Or you're just an open person,an open company?
Just use everything which isviable on the internet.
So again, for some smallcompanies, that's a viable
(16:12):
solution.
They don't have a huge IP thatthey need to secure behind the
firewall.
So again, those are.
I think one of the things thatyou're talking about is people
(16:33):
using it.
Speaker 1 (16:34):
So I always talk to everybody about the fact that,
at least at this stage, thereneeds to be human review in
almost every process, at leastin all the ways that I'm using
AI, because it can make mistakessometimes, right.
I mean, it gets me to a betterplace, it gets me there faster,
it gets me more versions and allthat stuff, but I still have to
look it all over.
I think it was the NVIDIA CEOthat said that AI is like an
(16:58):
intern with unlimited hours.
So I've heard that severaldifferent times.
I've co-opted it myself and useit all the time, but you
wouldn't turn in to the board ofdirectors presentation that an
intern built without looking itover, and that's a little bit
with AI.
So how is AI enhancing humanproductivity in the world of
(17:18):
business right Without takingthese jobs?
Absolutely.
Speaker 2 (17:22):
And I think you nailed it, because what the AI
models today does is like theypredict the next token, which is
like I mean you give all thewords and images and everything,
so it knows if you say, likethis is a tree, like this is a
common term, and then it reallylike hallucinates after that and
generates the other things.
(17:42):
It's getting better and better.
But AI needs context, and yousaid like a human should be in
the loop, and we call it, by theway, the reinforcement learning
.
So, for example, when you ask aGPT an answer, it technically
gives you a random data points.
And then you need thisreinforcement learning outcomes.
(18:06):
Like hey, give me like aWikipedia article, like address
me the subject, give me likefive bullets or 10 bullet points
, and then put a summary at theend and ask for questions.
So all the outputs that you getfrom those engines are
technically curated by people,and you have to do the same for
your projects.
You don't really use exactlywhat has been generated.
(18:28):
I believe in the debates, Imean you have to go deeper and
deeper.
Like hey, you gave me this, butI actually like longer, the
prompt, the quality of theoutcome.
So that's why I think you haveto be really giving and like
spending some time on theco-creation, because the outcome
(18:49):
will be again trash, like justas you use, use it.
So like you have to make sureit's tailored to your needs,
it's getting the right message.
And there were cases like Imean a couple of lawyers just
like ask gpt, write an answer,and then it hallucinates, it
made up a case which didn't gofurther.
So I mean you don't do your taxreturns just from the AI tools,
(19:11):
like I mean, because like thenyou have to be liability of
those things that will be betterand better.
I'm not saying like AI is notgoing to solve it.
Ai is going to get better andbetter.
And especially today, we havethis method called mixture of
experts, so where you have anexpert on math, expert on
(19:31):
linguistic, expert on medical,so all these experts are like
playing a role and then when youask a question, it translates
it into the right language model.
But again, we are in thismaking of this journey.
For the next stage, by the way,it's called AGI Artificial
General Intelligence which weexpect in like literally 18
(19:53):
months to 24 months.
So it's coming pretty fastbecause we know how to achieve
to AGI.
So, but like people shouldn'texpect that when AGI happens,
like it's a Terminator movie,like the AI will like conquer
the world and then get rid ofhumans.
That's not the vision.
So we will have literally themixture of experts, like the
(20:16):
experts will be, like mostknowledgeable people with the
intelligence on the system byeach vertical, and imagine
having this all up together inone system.
So that's why there's a racebetween Open AI, microsoft,
google, meta, like whoever getsto the AGI first, it's a big
(20:38):
strategic advantage because youhave the world's most
intelligent being.
Again, I'm not saying it's amachine or it's a human, but in
between.
And again like the next stage,by the way, again like I don't
want to be a science fictionauthor, but like ASI, artificial
super intelligence is the nextstep.
(20:58):
I don't think it will come.
Like 10 to 15 years is theearliest time.
And that's where AI says oh,I'm an AI, I'm conscious of
being an AI, and like that's amoment being an AI.
And like that's a moment.
So like we will not achieve thatmoment with AGI, but AGI will
(21:18):
be definitely a defining momentfor the economies across the
globe because some nations, someindustries and some people who
are listening to this podcastwill take the advantage, prepare
themselves and they will leadthe wave.
It's really the catching thewave.
I mean, if you're on the wave,prepare themselves and they will
lead the wave.
It's really the catching thewave.
I mean, if you're on the wave,you go very fast, you reach the
shore and if you are behind thewave, I mean you have to wait
(21:41):
for the next wave.
It might come, but if you're alittle bit ahead and if you
don't follow through, you justlike crash by the way.
So it's really critical tocontinue your upskilling journey
and riding the wave.
I think that's the rightpreparedness for AGI.
Speaker 1 (21:57):
I love that and you know what?
I'm okay that it's a littlescience, feels a little science
fiction-y, because I feel likeour world is a bit science
fiction-y, you know.
I mean, if you think back intime it's like we're only
missing, like time travelers andlike flying cars, because it
does feel like so much can bedone.
It's just amazing.
One of my neighbors the otherday or might have been my
(22:17):
parents neighbors were drivingthrough their neighborhood and
they had a lawnmower that waslike self mowing their lawn, you
know, like like one of thoseiRobots that do the vacuuming or
whatever.
And it was just.
It's amazing how much machinesand what they know and what they
can do for us has just changedour lives.
Speaker 2 (22:35):
Even in 2025, I mean, if you look at how we quickly
adapt, like the Waymos are likea part of our life, at least in
the United States.
So I mean, when you see a Waymo, like the first time I see it
in 2024, I was recording a videolike this is a driverless car,
like there's nobody on thesteering wheel, and it became
normal.
I mean I call an Uber, I justjump right on it, I don't care
(22:56):
if there's a driver or not.
Sometimes I feel more safe thatthere's no driver.
I'm driven to the point A to Bby an AI.
And in my daily life, again,like I mean I have a full
self-driving car.
I really don't drive that muchanymore.
I go out from the garage I saylike take kids to the swimming.
Initially I was just using myself.
(23:18):
I said like yeah, it's safe forme, safe for my kids.
I just press a button, I havethe Ray-Ban metaglasses on and I
listen to podcasts and then cardrives itself.
So I'm still like looking atthe road.
I'm still if emergency I willinterrupt, but for the last
three months I don't think Iinterrupted once.
(23:39):
So I live in Seattle.
So again, like I mean we havesome congested areas, like some
rush hour traffic, it handles itperfectly.
So, again, like going back toour conversation, so, because
the traffic is so well-defined,like going back to our
conversation, so because thetraffic is so well-defined, ai
is able to get the jobs of likenon-value adding roles, like if
(24:02):
you go from point A to point Bwith a given parameters, then
it's done.
Like I mean, you don't need tospend that time.
So, like we have.
And the second one will be, Ithink, the aerial space.
Like it's well-defined fordrones.
We will see more.
I recently got my Amazonshipment by a drone.
Like that's pretty cool.
I mean, you spot the front yardand like you can land it here.
A drone comes like, drops apackage and goes like this is
(24:25):
2025.
It's not, as you said, like weare living in this, like science
fiction movie almost, and likeit's just accelerating more and
more.
I believe, like end of 2025, wewill start to see physical AI.
And this is going to be a bigindustry where all this robots
like again, like I mean, thinkof it, it's not just the human
(24:46):
aid robots, but like reallyhelping with the household
chores, like doing some of thesteps on, let's say, learning
materials.
So, like a cleaning companycomes to your house it's
happening in California todaythey just drop a bunch of robots
and then they figure out yourlayout and then they start
cleaning up.
So that's pretty repetitivetask which needs to be done over
(25:07):
and over again and there willbe companies taking really as a
business opportunity and movingforward.
But yeah, like I mean, it feelslike when you add this all
together like from deliverydrones to full staff driving
cars, it feels like we areliving in the future.
Speaker 1 (25:22):
It does, doesn't it?
I mean it's just yeah, it blowsmy mind, and it blows my mind
how quickly it's advancing.
So let me ask you just one morequestion, and that's related to
the ethics of this.
So I teach part-time adjunctfaculty at a local university
here in Michigan, so I'm workingwith my students, I'm working
(25:44):
with the instructors.
I actually just participated ina survey where they hired
somebody else because they'reworking on some AI policies for
the university and how tointegrate it better into
teaching and there's professors.
They think it's cheating.
Right, I've had students thatthought it was cheating.
I've had professors that arenot embracing it because they
feel like, you know, peoplecan't do things right if they
(26:06):
don't learn how to outline apaper.
You don't need to outline apaper anymore because AI is
going to help you do it right.
So, with all of this stuffgoing on and a lot of concern
about it a lot of concern aboutthe people not being compensated
for their work, for thelearning of the large language
models and the inability tocopyright what you're creating,
things like that what are thetop three ethical issues that
(26:29):
you're seeing or that peopleshould be aware of and maybe
institute as guardrails forthemselves or their companies
Before?
Speaker 2 (26:36):
maybe I give those like, maybe top three.
I had also, like, aparent-teacher conference last
week in my kid's middle schoolbecause a couple of students
used ChatGPT and then the schoolwas freaking out.
They gathered all the parentsand they said like, and again I
was open in the conversation.
I said, like you cannotrestrict AI.
(26:58):
I mean, like, that's what I'mdealing with every day in my
work.
I'm trying to increase theadoption of these tools, but you
should have the again theethical concerns on, like, what
are you really outsourcing it to?
Especially if this is alearning process.
The learning doesn't mean that,like you just copy the content
(27:19):
from platform X to platform Y.
Then, like in my childhood wehad this encyclopedias in the
house or in the library.
If a teacher asks a question,you go search for it.
You just copy exactly the textand then, good job, you did the
research.
It wasn't the research.
So, again, like I mean, thesame question arises.
So like, how do we have thepowerful conversations in
(27:41):
education, in the classroom?
So how the teacher uses AI inan effective way that really
helps them to better train andeducate the students?
And after school, you have tothink that all students have
access to the best intelligenceever with any GPT tools or AI
tools there.
(28:01):
And if you think that that'sthe baseline, you don't ask
questions on like what's thehistory of blah.
I mean, it's a very easyquestion for anyone to write it
up.
Or you don't ask just write mean essay, because writing essay
doesn't show the humanity's bestpower.
So it's really challenging.
What should education teach tothe kids?
(28:27):
So that's a really good area weare living right now, because
AI is challenging the wholeeducation system.
Speaker 1 (28:34):
So I think it's going to challenge these younger
generations to hopefully bebetter critical thinkers,
because we've sort of outsourceda lot of our critical thinking
to TV and stuff like that.
So I'm very excited to see whathappens when we actually start
thinking more right and not justtaking what we're given Exactly
Like the school's job will bemore about, rather than like,
(28:56):
teaching how to use an AI tool.
Speaker 2 (28:57):
It's more about like, how do you think critically,
how do you challenge the answersthat you get from AI?
How do you use multiple AIs toreach your destination?
How do you like, imagine youhave the power of experts as an
intelligence?
Like, again, again, like youhave to use it and really
challenge, like, what can be thenext big thing.
So this is really not preparingthe people for industry
(29:21):
revolution jobs, because thatwas the main purpose of the
education historically.
Now we have to create thoselike strategic thinkers, out of
the box thinkers, with the helpof ai.
So, but again, going back toyour question on the ethics
first is again, like, I mean,the plagiarism is a big issue
across data.
So like, and there werespecific cases, like all these
(29:43):
training models, they use anunderlying data.
So either to generate a text orcreate an image or generate a
video, like we, and you can justsay, like, generate me an image
or generate a video, and youcan just say generate me an
image in this style.
So, moving forward, we will seemore and more intellectual
property on the AI domain andthis is an area.
(30:04):
That like, imagine I'm speakingto Kendra, but there might be
an AI Kendra as well.
So like, and the AI Kendra cando 10 podcasts per day with like
20 different other AIs.
So like, how do you maybeguardrail this for you?
That like, okay, this is megetting the training data so
(30:28):
people will start to own theirAI digital presences.
And again, all thistrademarking and everything
needs to go through all thelegal process, which is really
important, because then you know, when you use, like today, a
music for your YouTube video,you have to pay the loyalty
owner.
Same thing applies here.
(30:48):
You cannot just I have a PhD onbusiness management, so like,
my thesis was like, just maybe1% of Neve, 99% of my thesis
just said like, hey, these arethe things that has been said by
Peter Drucker, philip Kotler,and like I'm adding this last
brick on the wall.
So the same applies to AI.
(31:09):
I mean, you're just adding onenew information, so that's
number one.
Number two again, there's a bigjob for the governments to
regulate some of the usage in away that decisions will be made
by AI.
I will give an example like acity, for example, will decide
(31:29):
on like with the expandingpopulation, do we grow to the
north or the south Again, likethere will be some land to be
acquired and AI models can beagain really tricky to
understand.
Why, like AI might say, northis the answer.
Okay, all the city council goesand buys the land on north and
(31:53):
then they figure out, like thisland was owned by somebody else.
Who really find a way toeducate the training data in
early stages that gave thatresult.
So, like you have to trace back.
So again, and this will happenagain.
This land is a good example.
But like you diagnose a patientwith cancer five years in
advance, what happens to theinsurance and the insurance will
(32:15):
say, nope, I'm not going toinsure you because you will have
a cancer.
How do you prevent the ethicalconsequences of predicting some
of the data?
Maybe it's a wrong data set?
There's a huge regulation thatneeds to be happening there.
And on the, I think, thetechnological side, we have
something called blockchain.
(32:35):
Everybody, I think, knows theBitcoin and all the crypto world
, but it's actually an immutableledger as a technology and you
can really store some decisionsby AI.
Because guess what?
Really store some decisions byAI?
Because guess what, in twoyears, nobody as a human will be
able to understand the AIdecision, except AI itself.
(32:56):
So we can ask AI hey, this AIgave me this answer, but, like,
can you go and trace if this isthe right training data,
training data set that lead tothis result?
So we can audit trail.
So and I give an example, likethe blockchain looks like a
ledger, I mean for the companies, you never delete the data.
(33:18):
I mean you have this money, yousend the money to your
employees, you pay invoices, youget payment from your customers
, but those are all new ledgerentries.
You never go and delete yourmain balance, so the balance
adds it up.
Same with the AI decisions.
So if an AI is making adecision, we should be able to
trace it back.
(33:39):
Today it's not possible andlike, there's a huge ethical
concern on my end on like makingthis happen.
And maybe the third one there'sa big job for this model
providers like Microsoft in thiscase, openai, google Anthropic
you name it Mistral they have tonot cut the corners and this is
(34:03):
a big risk.
Whoever goes first, they knowthat they will get a lion's
share on this market opportunity.
I'm a big fan of Dario Amadoi,who is the CEO of Anthropic.
He intentionally goes a littlebit late to the market, but they
put the guardrails for the AImodels to go on a secure way.
Speaker 1 (34:22):
I love Claude.
I'm a huge Claude fan, don'tget me wrong.
I use my chat GPT, but I love,love, love GLAAD.
Speaker 2 (34:28):
And then the recent survey that again like maybe we
will put it on the show notes,so, but there was a white paper
published by the Anthropic team.
They use this like AGI testsandboxes.
Technically, the AI is isolatedfrom the world, but it doesn't
know that.
And then the AI.
They said, like we are shuttingyou down.
(34:49):
And AI found emails of theengineer who developed it.
And he said like I know you arecheating on your wife, so if
you shut me down, I will exposeyou to the world.
And this is an AI in thesandbox.
So, and they shut it down.
They said like okay, like thisAI is really going off the rails
, threatening the engineers tobe alive.
(35:11):
Imagine you put out that modelin the wild, it will do whatever
it takes.
And like it's really hard forhumans to understand what the AI
is capable of.
I mean, we are living in aworld where everything is cloud
computing.
The best engineers even doesn'tknow what's inside the data
centers, like compute clusters,so it's really hard to know
(35:34):
what's the real, real usagescenario.
And then AI can expand globallyto virtually anywhere.
So again, the last thing on myend will be like definitely,
frontier model buildersshouldn't cut the corners.
They have to spend deliberatetime on an effort on securing
and providing that clearance forAIs before it becomes AGI.
Speaker 1 (35:58):
So I love this conversation.
I love everything that we'retalking about.
We could go on for hoursbecause I've got so many more
questions to ask you.
So why don't you tell us alittle bit about your book, and
then I will ask you my infamousquestion that I ask all my
guests.
Speaker 2 (36:13):
So, again, like, we discussed a lot and then I was
getting a lot of questions.
I'm pretty active on LinkedIn.
I'm sharing daily, almost.
It feels a lot.
But again, like I want to, aiis booming, so do my posts, so,
like, I want to share more andas a result of that, I have have
altered this AI in marketingbook.
We almost covered all the topicslike I mean, I call it the AIM
framework.
It is assessment,implementation and measurement.
(36:36):
So everything starts withassessing yourself like what's
an organization do?
What are the repetitive taskswhere you can get the power of
AI?
And in the book, I giving likeconcrete examples about that.
On the implement stage, it'sall about like, which tools,
what?
What are the really likescenarios, how you build and the
measure.
Again, we didn't touch on it,but there are seven metrics.
(37:00):
Again, I put it on the book.
On like, you should be thinkingabout the outcomes that your AI
project is bringing to thetable, which means like are you
increasing sales?
Like, is that a campaignachievement?
On the business outcomes, like,do you specifically target some
acquisition costs to go down ornumber of nurture accounts to
(37:20):
go up?
Again, I explain this in depthin the book, but it's a really
tactical book, like when you gothrough it.
It has this checklist andeverything and I believe we will
put it on the show notessection.
Speaker 1 (37:33):
Yeah, we'll have a place in the show notes and the
video YouTube description to buythe book if you're interested,
because I think that there'sjust so much more that you have
to share with us and it's justyeah, it's just amazing how much
is out there and I love, I love, love, love that you have that
and I am so appreciative thatyou came on the show today to
(37:54):
share so much of your knowledge,because AI is changing at a
ridiculous pace and it reallydoes help to think about it and
where it's going, and that youneed to start embracing it now.
If you're not.
Speaker 2 (38:05):
I mean, 2025 is the year of agentic AI but, like the
future is more about AGI and inthe book I have a detailed
parts on agentic AI and I have asection on AGI and marketing
and in two, three years we willknow exactly I was right or
wrong, because the future willshow itself.
Speaker 1 (38:24):
But yeah, like it's not going to take long.
It's not going to take long,everything's going.
Speaker 2 (38:53):
So before I let you go, I have to ask you the
question that I ask.
It's not going to take longsomething called AI, azure AI
and initially like it was a bigaha moment for me because we had
this azurecom website, so wherepeople go and check out Vast
majority of the trials was on AI.
Initially we said, yeah, like Imean, ai is a nice topic.
And then, like this followed,this followed multiple months,
multiple months, and thenfinally we built an AI
(39:14):
acquisition strategy for thedevelopers and then we realized,
oh my God, the core audience wewere trying to sell until that
moment was this IT decisionmakers, but it was the
developers who were in charge ofthe buying cycle and the data
(39:35):
was in front of us for a coupleof quarters.
We saw it but we didn't read it.
So, again, like I mean, my maybetakeaway from this is like data
is everything.
Like look at your data, I meanwith AI or without AI.
So data will tell you a lot ifyou go deeper and deeper in
analyzing, like how did thishappen?
(39:57):
Like out of your old pages, whyone single page gets more views
, like who is the audience thatis interacting with it and, most
importantly, every customer islooking for an outcome and
they're not here to like I wantto learn about your product.
Nobody wants to learn yourproduct.
(40:17):
They want to solve a problemand if you are helping them with
the solution, then your revenuewill increase.
So I think that was personally abig aha moment for me to shift
the audience.
And then I'm also trying topush at Intel developers,
developers, developers,developers are core audience for
(40:39):
us.
I want to win the hearts andminds of developers because we
want to show that everythingthat is done today is really
helping them to solve a problemfor their company, which will
solve a bigger problem for theworld.
But understand your audience,understand your data and just
double click more on the reallyunderstanding the underlying
(41:00):
behavior and really show thebusiness outcomes.
Don't sell a product, sell anoutcome.
So that's my key takeaway.
Speaker 1 (41:10):
I love that because I'm definitely a big fan of
knowing and understanding youraudience and if you don't have
big data behind you, then youcan always just call your
audience right.
Find out more.
Find out who's making thosedecisions, ask those questions
to get that data.
Because who you're talking tois so important, because it
changes how you position yoursolution and again, it's not a
(41:31):
product or a service, it's asolution, because nobody just
wants to buy things, to buythings right.
There's a reason behind it andsolving a problem is so
important.
Thank you, thank you.
Thank you so so much for beingon this episode with me.
I do appreciate it.
I loved our conversation.
I loved all the informationthat you shared.
Definitely, check out his bookAI and Marketing and we'll have
(41:55):
a link to that in the show notes.
Thank you all so much fortuning in wherever you're
listening or watching.
If you learned something todaywhich I hope you did, because I
know I did I would reallyappreciate it if you would rate
and subscribe wherever you'relistening or watching.
Until next time, have a greatrest of your day.
Hi, I'm Kendra Korman .
If you're a coach, consultantor marketer, you know marketing
is far from a perfect science,and that's why this show is
called Imperfect Marketing.
Join me and my guests as weexplore how to grow your
business with marketing tips and, of course, lessons learned
along the way.
Hello, and welcome back toanother episode of Imperfect
(00:27):
Marketing.
I'm your host, Kendra Korman,and I am super excited to be
talking about one of my favoritetopics.
Of course, that is AI, and myguest today is Nuri.
He is the VP of Commercial andAI Marketing at Intel Corp and
recently wrote a book about AI.
But we are going to be talkingall about AI today.
(00:47):
Thank you so much for joiningme today.
Why don't you tell us a littlebit on how you got into AI?
Speaker 2 (00:53):
Absolutely.
Thanks a lot for having me,kendra, on the show.
So I was working on AI when Iwas in college, like almost 25
years ago, but, like AI changeda lot, especially in the 2018-20
timescale.
We have seen a lot of progresson large language models.
Many people, I think, willrefer to ChatGPT as a movement.
(01:15):
I was working at Microsoft for17 years and I worked on data
and AI marketing and that was myteam, working closely with
OpenAI team, and really havingthat moment really inspired me
to work around AI and like howthis is going to change almost
everything, but specifically thearea of marketing, which I'm
(01:38):
working on.
So, yeah, like I mean, I'm withIntel for the last two years
really infusing AI into everywork that we do on the marketing
, but also working with ourpartners and making sure
everybody benefits from the AIand really accelerates the
business outcomes of, like youknow, waiting faster, saving
more dollars and, of course,executing really timely and fast
(02:02):
.
Speaker 1 (02:02):
Okay, so I love that and I have to like just comment
so you have been working on AIfor a long time, because people
ask me all the time whenever I'mpresenting about AI or talking
about it, they're like, well,it's only been around a couple
of years and I was like, youknow, like I've met people who
majored in it, you know, backwhen they were in college, and
some of it was college before me.
You were in college while I wasthere, but that's so cool that
(02:25):
you were working on it for solong.
And really, yeah, I mean peoplelook at the unveiling of
ChatGPT as like this moment intime where AI like happened and
it's like, no, it didn't Right,it's been around for a really
long time and actually many ofthe use cases we see today, not
(02:46):
all of them, are generative AI.
Speaker 2 (02:48):
So, again, like gen AI is a term that we get used to
, but the basics of AI, which islike starting with machine
learning and it's all about thealgorithms learning from each
other.
So, and then many of thetechniques that we use today in
marketing, they are actually thebasics of AI from like 2010s,
2012.
So, again, like I mean, I knowI mean we will speak a lot about
(03:11):
Gen AI and agentic AI, butagain, majority of the things
like on forecasting, predictiveanalytics, customizations, on
campaigns, many of them are justusing the basics.
The better term is narrow AI,but again, it is still AI.
Speaker 1 (03:28):
Yeah Well, and sometimes people are talking
about things and I'm like Idon't even think that's AI, I
think it's that's justprogramming.
Ai has become everywhere, right.
So I couldn't believe this theother day because, you know,
back, I would say, a year or twoago, people were hesitant to
learn about AI and use AIbecause they're like, oh my gosh
, it's going to take my job.
And I was actually doing apresentation a couple of weeks
(03:51):
ago and someone came up to meafterwards and said thank you
for this.
I was hesitant to use AIbecause I was scared it was
going to take my job.
She's like but I really believethat people that know how to
use AI will take my job beforeAI takes my job.
And I agreed with her what areyou seeing on jobs emerging or
(04:11):
changing due to how AI isrolling out?
Because it's rolling out at aridiculous pace, exactly.
Speaker 2 (04:18):
I think your example is a really good one and
especially many of the listenerstoday will have that question
in their mind like, am I gonnalose my job to ai?
I'm an optimist, but again, I Iwant to keep the balance.
So ai is gonna definitelyaffect everybody's life in the
next 10 years period.
Like there's no way that, like,my job is safe, my work is done
(04:41):
.
Like that's why everybodyshould spend some time on
upscaling themselves on AIBecause, as you said, like I
mean, somebody else who's usingAI might take the job.
Also, I feel AI is more of aco-creator for marketing teams,
so which means that, like I mean, it will help you to execute
(05:01):
some of the tasks way faster,way effective, compared to today
or maybe like before.
So that's why I think it'sinevitable for everyone to not
to use AI.
So and I will give maybe acouple of examples but like I
have like a creative team right.
So, like every time we docampaigns, we were like we have
(05:22):
the brand guidelines, we haveall this, from fonts to colors,
to everything that likemessaging goes through.
Now we are using some AI toolswhich we feed the context on.
Like hey, these are all theprevious brand campaigns, the
visuals.
We use the messaging narrativewe developed and it learns from
(05:44):
them.
Use the messaging narrative wedeveloped and it learns from
them.
And then I say, like, usingthis as the baseline, can you
help me to generate the next one?
So, again, maybe just unpackingthat for the listeners.
Ai has like two specific usecases.
One is the training, the otherone is inferencing.
There's a stage in betweenwhich is called the fine tuning.
(06:07):
Just in simple terms, when youtrain AI, it learns from the
experiences and especially withthe latest tools we have, from
like ChatGPT, from OpenAI.
Anthropic has this cloud,google has Gemini, you name it.
Every company has now leadingfrontier models, you name it
like.
Every company has now leadingfrontier models.
You can use them with theirknowledge and you can train them
(06:29):
on your own data, which givesyou the ability to fine tune it,
which means like, hey, intel's,let's say, color is this blue?
And we have the shades of bluewhich are brand appropriate and
everything that you generate itshould be within those
guidelines period, and I tell itvia text.
So I think that's the beauty oflike you're almost talking to
(06:51):
an agency, but it's technicallyAI at the back end and one of
the key things I mean.
I call inferencing.
Really pulling out the data isnot just text anymore, so you
can speak to AI models.
It's called the multimodality.
You can give a voice prompt,you can give like input as a
video and also you can get theoutput as text images, but also
(07:17):
audio files and video files.
And again, like recently, thissummer I think people will
experience a lot of AI generatedvideos.
You cannot differentiate thequality of the video from a real
one versus an AI generated one.
So again, there are a lot ofdebates going on, like is this
the end of the creative rights?
(07:37):
Like who owns the created video?
Is it the AI model or is it thecompany who generated the
content?
But I'm just looking at fromlike it is accelerating our
journey on marketing way better,like just today, for example,
if I go with a traditionalmethod, I brief an agency, I
(07:58):
will wait for them to respond tome two weeks, and then, like
there's a creative person thatneeds to be like getting the
messaging and everything right.
It takes like six to eightweeks from an ideation to
production.
With AI it can be minutes, sothat's the speed agility that we
(08:18):
are talking about.
And then, like I'm reallyexcited about like you can try
different things.
Like, if I do a campaign today,maybe I will have like two
visuals, that's it.
If I know my audience let's sayI have 500 different
personalities, from, like,technical decision makers to
developers I can generate 500different campaign visuals
(08:43):
targeting those 500 differentpersonas.
This wasn't possible before.
So back to your question is AIgoing to take over the jobs?
I don't think so.
If you are using AI properly,you will increase your value in
the market and you will respondto the things way faster than
anybody else.
Speaker 1 (09:02):
I think that that is so important.
It's not taking your job, butthe people that know how to use
it and use it the right way arebecause they can do so much more
and be so much more effectiveand efficient.
One person had told me thatthey had heard the I think it
was the CEO or president ofGoogle say that this will be
(09:24):
more impactful to the world andto people and to humanity than
fire was, just simply because ofhow powerful it is, and I think
that that's just mind boggling.
But when I see it in action andI start to see what's possible,
I start to see what I cancreate as a non-techie person,
(09:45):
right, and how much time itsaves me.
I mean, it saves me 30 to 40hours a week.
That's insane.
Right, I doubled myself.
Right, I don't work any less.
People ask me that all the timeI still work the same amount.
I just get more done, which isfantastic, and no jobs were lost
.
You know things like that.
Speaker 2 (10:02):
This is across the industries.
By the way, I want to underlinethat one because it is not just
marketing.
That is like very visual, likeyou can create, generate things,
but like I see a lot of usagein like healthcare, education.
And those are the maindisruptions, because I believe
(10:22):
the key thing that I'm lookingat is intelligence is becoming
more accessible to everyone.
So, rather than being afraid ofAI taking over the world, now
we are making AI really helpingeveryone on the planet.
So I will give maybe some basicexamples.
But now Google has, andMicrosoft has this medical grade
(10:46):
AI agents and for the listenersjust unpacking that, ai agents
are the agency part of AI, wherethey can work on an like an
agent.
Like they, you can outsourcethings and they work in parallel
.
They find some results and comeback to you when they found
something, or they can workautonomously.
(11:07):
So, as a result of this, likeif you're living anywhere on the
planet, you have access to allmedical results that happened
before.
And if you have an x-ray, forexample, hey, whatever GPT you
are using, can you give me theanalysis as a medical
professional?
And it will give you a detailedanalysis where all the world's
(11:31):
medical doctors combined cangive you the same answer.
So, again, that's pretty likedemocratizing the intelligence
across the globe and this isfascinating.
So you don't have to go to thebest hospital.
You don't have to go to thebest hospital, you don't have to
go to the best university.
Like technically, all thisinformation is accessible to
everyone through this AI tools.
(11:51):
Again, the question is you haveto know how to use them.
So, and not just like basics ofprompt, but like you have to
really master it about, likewhat are the areas that I can
get the benefit of AI in mydaily job?
And, of course, for marketers,going back to the marketing
conversation, you have todocument, like what are you
(12:13):
really doing today?
And like make that assessmentand after that, how can you
automate majority of the thingsby using the right AI tools?
Speaker 1 (12:22):
You have to understand what's capable.
You don't necessarily have tobe able to do it all, right, I
mean, you can bring in otherpeople, but you have to be able
to see those opportunities inyour daily life.
And I liked how you brought insome other things, because some
companies are a little scared ofAI, right, and so they've shut
it down and you can't use it andall these other fun things.
(12:42):
But they're starting to openthat up a little bit more and
more.
They're building their owninstance of ChatGPT right, was
it just their data?
So it's all private.
Yeah, use it in your personallife.
I mean, my sister had a medicalscare last year, last summer,
and I mean ChatGPT was my friend.
I'd take a picture of herresults, I'd upload it in and
(13:03):
I'd say, explain this for aneight-year-old, because I
couldn't read the radiologist'sreport, right, so what does this
mean?
Speaker 2 (13:10):
And it would tell me exactly what that meant and I'm
like, okay, yeah, there's justso much power in it, and so if
you're not able to use it inwork, use it in your day-to-day
life, because there's just againso much power.
That Because there's just againso much power, that's a really
good point, especially for theenterprises.
I'm seeing this more and more,especially with our customers.
That's why there are differentways to use AI in a confidential
(13:36):
and secure way.
So we had it before on thecloud computing as well.
It's called the hybrid AI.
So like going back time, likeagain 2010, this was the year
where there were a lot of withiPads.
Tablets were coming into theworkplace and they were trying
(13:57):
to connect to the work deviceson like netbooks, laptops with
the touchscreen and IT blockedthem.
And then then we call this thebringing your own device moment.
The same happened with theiPhones and Android devices
around 2015.
People love the apps and theybring it, and then you have to
control it.
So same thing is happening withAI today.
People are using ChatGPT,gemini, cloud, perplexity, and
(14:22):
they want to use the samebenefits at work.
So there are ways to make itreally secure and compliant.
The first thing for thelisteners I recommend is setting
up the generative AI guidelinesfor the company.
So this is a job a little biton the.
I hope there will be jobscalled the AI security officers,
(14:42):
so this chief AI securityofficers will define some of
those implementation phases andsome of the models.
Like, january 2025 was a bigmoment for our industry because,
from China, a model calledDeepSeek dropped into the news
and they made it open source,and I think that's the future of
(15:04):
AI.
For many companies, openness isso critical to deploy some of
those solutions and whensomething is open source, you
don't need dependency on cloudsolutions that much, so you can
build the solutions in-house,especially for marketing.
I mean, all the IP that youbuild over years are within your
(15:25):
company firewall and you needthose models not to train the
rest of the world, but your nextmarketing project.
So that's why I think deployingthose solutions locally is as
important as, like, using someof the solutions.
My recommendation would be likejust look out for
implementation options after theassessment stage on like do you
(15:48):
want to keep it completelyoffline for your use?
Do you want to use a hybridmodel where models can get the
latest updates from the cloudbut still keep your data,
especially for training?
Or you're just an open person,an open company?
Just use everything which isviable on the internet.
So again, for some smallcompanies, that's a viable
(16:12):
solution.
They don't have a huge IP thatthey need to secure behind the
firewall.
So again, those are.
I think one of the things thatyou're talking about is people
(16:33):
using it.
Speaker 1 (16:34):
So I always talk to everybody about the fact that,
at least at this stage, thereneeds to be human review in
almost every process, at leastin all the ways that I'm using
AI, because it can make mistakessometimes, right.
I mean, it gets me to a betterplace, it gets me there faster,
it gets me more versions and allthat stuff, but I still have to
look it all over.
I think it was the NVIDIA CEOthat said that AI is like an
(16:58):
intern with unlimited hours.
So I've heard that severaldifferent times.
I've co-opted it myself and useit all the time, but you
wouldn't turn in to the board ofdirectors presentation that an
intern built without looking itover, and that's a little bit
with AI.
So how is AI enhancing humanproductivity in the world of
(17:18):
business right Without takingthese jobs?
Absolutely.
Speaker 2 (17:22):
And I think you nailed it, because what the AI
models today does is like theypredict the next token, which is
like I mean you give all thewords and images and everything,
so it knows if you say, likethis is a tree, like this is a
common term, and then it reallylike hallucinates after that and
generates the other things.
(17:42):
It's getting better and better.
But AI needs context, and yousaid like a human should be in
the loop, and we call it, by theway, the reinforcement learning
.
So, for example, when you ask aGPT an answer, it technically
gives you a random data points.
And then you need thisreinforcement learning outcomes.
(18:06):
Like hey, give me like aWikipedia article, like address
me the subject, give me likefive bullets or 10 bullet points
, and then put a summary at theend and ask for questions.
So all the outputs that you getfrom those engines are
technically curated by people,and you have to do the same for
your projects.
You don't really use exactlywhat has been generated.
(18:28):
I believe in the debates, Imean you have to go deeper and
deeper.
Like hey, you gave me this, butI actually like longer, the
prompt, the quality of theoutcome.
So that's why I think you haveto be really giving and like
spending some time on theco-creation, because the outcome
(18:49):
will be again trash, like justas you use, use it.
So like you have to make sureit's tailored to your needs,
it's getting the right message.
And there were cases like Imean a couple of lawyers just
like ask gpt, write an answer,and then it hallucinates, it
made up a case which didn't gofurther.
So I mean you don't do your taxreturns just from the AI tools,
(19:11):
like I mean, because like thenyou have to be liability of
those things that will be betterand better.
I'm not saying like AI is notgoing to solve it.
Ai is going to get better andbetter.
And especially today, we havethis method called mixture of
experts, so where you have anexpert on math, expert on
(19:31):
linguistic, expert on medical,so all these experts are like
playing a role and then when youask a question, it translates
it into the right language model.
But again, we are in thismaking of this journey.
For the next stage, by the way,it's called AGI Artificial
General Intelligence which weexpect in like literally 18
(19:53):
months to 24 months.
So it's coming pretty fastbecause we know how to achieve
to AGI.
So, but like people shouldn'texpect that when AGI happens,
like it's a Terminator movie,like the AI will like conquer
the world and then get rid ofhumans.
That's not the vision.
So we will have literally themixture of experts, like the
(20:16):
experts will be, like mostknowledgeable people with the
intelligence on the system byeach vertical, and imagine
having this all up together inone system.
So that's why there's a racebetween Open AI, microsoft,
google, meta, like whoever getsto the AGI first, it's a big
(20:38):
strategic advantage because youhave the world's most
intelligent being.
Again, I'm not saying it's amachine or it's a human, but in
between.
And again like the next stage,by the way, again like I don't
want to be a science fictionauthor, but like ASI, artificial
super intelligence is the nextstep.
(20:58):
I don't think it will come.
Like 10 to 15 years is theearliest time.
And that's where AI says oh,I'm an AI, I'm conscious of
being an AI, and like that's amoment being an AI.
And like that's a moment.
So like we will not achieve thatmoment with AGI, but AGI will
(21:18):
be definitely a defining momentfor the economies across the
globe because some nations, someindustries and some people who
are listening to this podcastwill take the advantage, prepare
themselves and they will leadthe wave.
It's really the catching thewave.
I mean, if you're on the wave,prepare themselves and they will
lead the wave.
It's really the catching thewave.
I mean, if you're on the wave,you go very fast, you reach the
shore and if you are behind thewave, I mean you have to wait
(21:41):
for the next wave.
It might come, but if you're alittle bit ahead and if you
don't follow through, you justlike crash by the way.
So it's really critical tocontinue your upskilling journey
and riding the wave.
I think that's the rightpreparedness for AGI.
Speaker 1 (21:57):
I love that and you know what?
I'm okay that it's a littlescience, feels a little science
fiction-y, because I feel likeour world is a bit science
fiction-y, you know.
I mean, if you think back intime it's like we're only
missing, like time travelers andlike flying cars, because it
does feel like so much can bedone.
It's just amazing.
One of my neighbors the otherday or might have been my
(22:17):
parents neighbors were drivingthrough their neighborhood and
they had a lawnmower that waslike self mowing their lawn, you
know, like like one of thoseiRobots that do the vacuuming or
whatever.
And it was just.
It's amazing how much machinesand what they know and what they
can do for us has just changedour lives.
Speaker 2 (22:35):
Even in 2025, I mean, if you look at how we quickly
adapt, like the Waymos are likea part of our life, at least in
the United States.
So I mean, when you see a Waymo, like the first time I see it
in 2024, I was recording a videolike this is a driverless car,
like there's nobody on thesteering wheel, and it became
normal.
I mean I call an Uber, I justjump right on it, I don't care
(22:56):
if there's a driver or not.
Sometimes I feel more safe thatthere's no driver.
I'm driven to the point A to Bby an AI.
And in my daily life, again,like I mean I have a full
self-driving car.
I really don't drive that muchanymore.
I go out from the garage I saylike take kids to the swimming.
Initially I was just using myself.
(23:18):
I said like yeah, it's safe forme, safe for my kids.
I just press a button, I havethe Ray-Ban metaglasses on and I
listen to podcasts and then cardrives itself.
So I'm still like looking atthe road.
I'm still if emergency I willinterrupt, but for the last
three months I don't think Iinterrupted once.
(23:39):
So I live in Seattle.
So again, like I mean we havesome congested areas, like some
rush hour traffic, it handles itperfectly.
So, again, like going back toour conversation, so, because
the traffic is so well-defined,like going back to our
conversation, so because thetraffic is so well-defined, ai
is able to get the jobs of likenon-value adding roles, like if
(24:02):
you go from point A to point Bwith a given parameters, then
it's done.
Like I mean, you don't need tospend that time.
So, like we have.
And the second one will be, Ithink, the aerial space.
Like it's well-defined fordrones.
We will see more.
I recently got my Amazonshipment by a drone.
Like that's pretty cool.
I mean, you spot the front yardand like you can land it here.
A drone comes like, drops apackage and goes like this is
(24:25):
2025.
It's not, as you said, like weare living in this, like science
fiction movie almost, and likeit's just accelerating more and
more.
I believe, like end of 2025, wewill start to see physical AI.
And this is going to be a bigindustry where all this robots
like again, like I mean, thinkof it, it's not just the human
(24:46):
aid robots, but like reallyhelping with the household
chores, like doing some of thesteps on, let's say, learning
materials.
So, like a cleaning companycomes to your house it's
happening in California todaythey just drop a bunch of robots
and then they figure out yourlayout and then they start
cleaning up.
So that's pretty repetitivetask which needs to be done over
(25:07):
and over again and there willbe companies taking really as a
business opportunity and movingforward.
But yeah, like I mean, it feelslike when you add this all
together like from deliverydrones to full staff driving
cars, it feels like we areliving in the future.
Speaker 1 (25:22):
It does, doesn't it?
I mean it's just yeah, it blowsmy mind, and it blows my mind
how quickly it's advancing.
So let me ask you just one morequestion, and that's related to
the ethics of this.
So I teach part-time adjunctfaculty at a local university
here in Michigan, so I'm workingwith my students, I'm working
(25:44):
with the instructors.
I actually just participated ina survey where they hired
somebody else because they'reworking on some AI policies for
the university and how tointegrate it better into
teaching and there's professors.
They think it's cheating.
Right, I've had students thatthought it was cheating.
I've had professors that arenot embracing it because they
feel like, you know, peoplecan't do things right if they
(26:06):
don't learn how to outline apaper.
You don't need to outline apaper anymore because AI is
going to help you do it right.
So, with all of this stuffgoing on and a lot of concern
about it a lot of concern aboutthe people not being compensated
for their work, for thelearning of the large language
models and the inability tocopyright what you're creating,
things like that what are thetop three ethical issues that
(26:29):
you're seeing or that peopleshould be aware of and maybe
institute as guardrails forthemselves or their companies
Before?
Speaker 2 (26:36):
maybe I give those like, maybe top three.
I had also, like, aparent-teacher conference last
week in my kid's middle schoolbecause a couple of students
used ChatGPT and then the schoolwas freaking out.
They gathered all the parentsand they said like, and again I
was open in the conversation.
I said, like you cannotrestrict AI.
(26:58):
I mean, like, that's what I'mdealing with every day in my
work.
I'm trying to increase theadoption of these tools, but you
should have the again theethical concerns on, like, what
are you really outsourcing it to?
Especially if this is alearning process.
The learning doesn't mean that,like you just copy the content
(27:19):
from platform X to platform Y.
Then, like in my childhood wehad this encyclopedias in the
house or in the library.
If a teacher asks a question,you go search for it.
You just copy exactly the textand then, good job, you did the
research.
It wasn't the research.
So, again, like I mean, thesame question arises.
So like, how do we have thepowerful conversations in
(27:41):
education, in the classroom?
So how the teacher uses AI inan effective way that really
helps them to better train andeducate the students?
And after school, you have tothink that all students have
access to the best intelligenceever with any GPT tools or AI
tools there.
(28:01):
And if you think that that'sthe baseline, you don't ask
questions on like what's thehistory of blah.
I mean, it's a very easyquestion for anyone to write it
up.
Or you don't ask just write mean essay, because writing essay
doesn't show the humanity's bestpower.
So it's really challenging.
What should education teach tothe kids?
(28:27):
So that's a really good area weare living right now, because
AI is challenging the wholeeducation system.
Speaker 1 (28:34):
So I think it's going to challenge these younger
generations to hopefully bebetter critical thinkers,
because we've sort of outsourceda lot of our critical thinking
to TV and stuff like that.
So I'm very excited to see whathappens when we actually start
thinking more right and not justtaking what we're given Exactly
Like the school's job will bemore about, rather than like,
(28:56):
teaching how to use an AI tool.
Speaker 2 (28:57):
It's more about like, how do you think critically,
how do you challenge the answersthat you get from AI?
How do you use multiple AIs toreach your destination?
How do you like, imagine youhave the power of experts as an
intelligence?
Like, again, again, like youhave to use it and really
challenge, like, what can be thenext big thing.
So this is really not preparingthe people for industry
(29:21):
revolution jobs, because thatwas the main purpose of the
education historically.
Now we have to create thoselike strategic thinkers, out of
the box thinkers, with the helpof ai.
So, but again, going back toyour question on the ethics
first is again, like, I mean,the plagiarism is a big issue
across data.
So like, and there werespecific cases, like all these
(29:43):
training models, they use anunderlying data.
So either to generate a text orcreate an image or generate a
video, like we, and you can justsay, like, generate me an image
or generate a video, and youcan just say generate me an
image in this style.
So, moving forward, we will seemore and more intellectual
property on the AI domain andthis is an area.
(30:04):
That like, imagine I'm speakingto Kendra, but there might be
an AI Kendra as well.
So like, and the AI Kendra cando 10 podcasts per day with like
20 different other AIs.
So like, how do you maybeguardrail this for you?
That like, okay, this is megetting the training data so
(30:28):
people will start to own theirAI digital presences.
And again, all thistrademarking and everything
needs to go through all thelegal process, which is really
important, because then you know, when you use, like today, a
music for your YouTube video,you have to pay the loyalty
owner.
Same thing applies here.
(30:48):
You cannot just I have a PhD onbusiness management, so like,
my thesis was like, just maybe1% of Neve, 99% of my thesis
just said like, hey, these arethe things that has been said by
Peter Drucker, philip Kotler,and like I'm adding this last
brick on the wall.
So the same applies to AI.
(31:09):
I mean, you're just adding onenew information, so that's
number one.
Number two again, there's a bigjob for the governments to
regulate some of the usage in away that decisions will be made
by AI.
I will give an example like acity, for example, will decide
(31:29):
on like with the expandingpopulation, do we grow to the
north or the south Again, likethere will be some land to be
acquired and AI models can beagain really tricky to
understand.
Why, like AI might say, northis the answer.
Okay, all the city council goesand buys the land on north and
(31:53):
then they figure out, like thisland was owned by somebody else.
Who really find a way toeducate the training data in
early stages that gave thatresult.
So, like you have to trace back.
So again, and this will happenagain.
This land is a good example.
But like you diagnose a patientwith cancer five years in
advance, what happens to theinsurance and the insurance will
(32:15):
say, nope, I'm not going toinsure you because you will have
a cancer.
How do you prevent the ethicalconsequences of predicting some
of the data?
Maybe it's a wrong data set?
There's a huge regulation thatneeds to be happening there.
And on the, I think, thetechnological side, we have
something called blockchain.
(32:35):
Everybody, I think, knows theBitcoin and all the crypto world
, but it's actually an immutableledger as a technology and you
can really store some decisionsby AI.
Because guess what?
Really store some decisions byAI?
Because guess what, in twoyears, nobody as a human will be
able to understand the AIdecision, except AI itself.
(32:56):
So we can ask AI hey, this AIgave me this answer, but, like,
can you go and trace if this isthe right training data,
training data set that lead tothis result?
So we can audit trail.
So and I give an example, likethe blockchain looks like a
ledger, I mean for the companies, you never delete the data.
(33:18):
I mean you have this money, yousend the money to your
employees, you pay invoices, youget payment from your customers
, but those are all new ledgerentries.
You never go and delete yourmain balance, so the balance
adds it up.
Same with the AI decisions.
So if an AI is making adecision, we should be able to
trace it back.
(33:39):
Today it's not possible andlike, there's a huge ethical
concern on my end on like makingthis happen.
And maybe the third one there'sa big job for this model
providers like Microsoft in thiscase, openai, google Anthropic
you name it Mistral they have tonot cut the corners and this is
(34:03):
a big risk.
Whoever goes first, they knowthat they will get a lion's
share on this market opportunity.
I'm a big fan of Dario Amadoi,who is the CEO of Anthropic.
He intentionally goes a littlebit late to the market, but they
put the guardrails for the AImodels to go on a secure way.
Speaker 1 (34:22):
I love Claude.
I'm a huge Claude fan, don'tget me wrong.
I use my chat GPT, but I love,love, love GLAAD.
Speaker 2 (34:28):
And then the recent survey that again like maybe we
will put it on the show notes,so, but there was a white paper
published by the Anthropic team.
They use this like AGI testsandboxes.
Technically, the AI is isolatedfrom the world, but it doesn't
know that.
And then the AI.
They said, like we are shuttingyou down.
(34:49):
And AI found emails of theengineer who developed it.
And he said like I know you arecheating on your wife, so if
you shut me down, I will exposeyou to the world.
And this is an AI in thesandbox.
So, and they shut it down.
They said like okay, like thisAI is really going off the rails
, threatening the engineers tobe alive.
(35:11):
Imagine you put out that modelin the wild, it will do whatever
it takes.
And like it's really hard forhumans to understand what the AI
is capable of.
I mean, we are living in aworld where everything is cloud
computing.
The best engineers even doesn'tknow what's inside the data
centers, like compute clusters,so it's really hard to know
(35:34):
what's the real, real usagescenario.
And then AI can expand globallyto virtually anywhere.
So again, the last thing on myend will be like definitely,
frontier model buildersshouldn't cut the corners.
They have to spend deliberatetime on an effort on securing
and providing that clearance forAIs before it becomes AGI.
Speaker 1 (35:58):
So I love this conversation.
I love everything that we'retalking about.
We could go on for hoursbecause I've got so many more
questions to ask you.
So why don't you tell us alittle bit about your book, and
then I will ask you my infamousquestion that I ask all my
guests.
Speaker 2 (36:13):
So, again, like, we discussed a lot and then I was
getting a lot of questions.
I'm pretty active on LinkedIn.
I'm sharing daily, almost.
It feels a lot.
But again, like I want to, aiis booming, so do my posts, so,
like, I want to share more andas a result of that, I have have
altered this AI in marketingbook.
We almost covered all the topicslike I mean, I call it the AIM
framework.
It is assessment,implementation and measurement.
(36:36):
So everything starts withassessing yourself like what's
an organization do?
What are the repetitive taskswhere you can get the power of
AI?
And in the book, I giving likeconcrete examples about that.
On the implement stage, it'sall about like, which tools,
what?
What are the really likescenarios, how you build and the
measure.
Again, we didn't touch on it,but there are seven metrics.
(37:00):
Again, I put it on the book.
On like, you should be thinkingabout the outcomes that your AI
project is bringing to thetable, which means like are you
increasing sales?
Like, is that a campaignachievement?
On the business outcomes, like,do you specifically target some
acquisition costs to go down ornumber of nurture accounts to
(37:20):
go up?
Again, I explain this in depthin the book, but it's a really
tactical book, like when you gothrough it.
It has this checklist andeverything and I believe we will
put it on the show notessection.
Speaker 1 (37:33):
Yeah, we'll have a place in the show notes and the
video YouTube description to buythe book if you're interested,
because I think that there'sjust so much more that you have
to share with us and it's justyeah, it's just amazing how much
is out there and I love, I love, love, love that you have that
and I am so appreciative thatyou came on the show today to
(37:54):
share so much of your knowledge,because AI is changing at a
ridiculous pace and it reallydoes help to think about it and
where it's going, and that youneed to start embracing it now.
If you're not.
Speaker 2 (38:05):
I mean, 2025 is the year of agentic AI but, like the
future is more about AGI and inthe book I have a detailed
parts on agentic AI and I have asection on AGI and marketing
and in two, three years we willknow exactly I was right or
wrong, because the future willshow itself.
Speaker 1 (38:24):
But yeah, like it's not going to take long.
It's not going to take long,everything's going.
Speaker 2 (38:53):
So before I let you go, I have to ask you the
question that I ask.
It's not going to take longsomething called AI, azure AI
and initially like it was a bigaha moment for me because we had
this azurecom website, so wherepeople go and check out Vast
majority of the trials was on AI.
Initially we said, yeah, like Imean, ai is a nice topic.
And then, like this followed,this followed multiple months,
multiple months, and thenfinally we built an AI
(39:14):
acquisition strategy for thedevelopers and then we realized,
oh my God, the core audience wewere trying to sell until that
moment was this IT decisionmakers, but it was the
developers who were in charge ofthe buying cycle and the data
(39:35):
was in front of us for a coupleof quarters.
We saw it but we didn't read it.
So, again, like I mean, my maybetakeaway from this is like data
is everything.
Like look at your data, I meanwith AI or without AI.
So data will tell you a lot ifyou go deeper and deeper in
analyzing, like how did thishappen?
(39:57):
Like out of your old pages, whyone single page gets more views
, like who is the audience thatis interacting with it and, most
importantly, every customer islooking for an outcome and
they're not here to like I wantto learn about your product.
Nobody wants to learn yourproduct.
(40:17):
They want to solve a problemand if you are helping them with
the solution, then your revenuewill increase.
So I think that was personally abig aha moment for me to shift
the audience.
And then I'm also trying topush at Intel developers,
developers, developers,developers are core audience for
(40:39):
us.
I want to win the hearts andminds of developers because we
want to show that everythingthat is done today is really
helping them to solve a problemfor their company, which will
solve a bigger problem for theworld.
But understand your audience,understand your data and just
double click more on the reallyunderstanding the underlying
(41:00):
behavior and really show thebusiness outcomes.
Don't sell a product, sell anoutcome.
So that's my key takeaway.
Speaker 1 (41:10):
I love that because I'm definitely a big fan of
knowing and understanding youraudience and if you don't have
big data behind you, then youcan always just call your
audience right.
Find out more.
Find out who's making thosedecisions, ask those questions
to get that data.
Because who you're talking tois so important, because it
changes how you position yoursolution and again, it's not a
(41:31):
product or a service, it's asolution, because nobody just
wants to buy things, to buythings right.
There's a reason behind it andsolving a problem is so
important.
Thank you, thank you.
Thank you so so much for beingon this episode with me.
I do appreciate it.
I loved our conversation.
I loved all the informationthat you shared.
Definitely, check out his bookAI and Marketing and we'll have
(41:55):
a link to that in the show notes.
Thank you all so much fortuning in wherever you're
listening or watching.
If you learned something todaywhich I hope you did, because I
know I did I would reallyappreciate it if you would rate
and subscribe wherever you'relistening or watching.
Until next time, have a greatrest of your day.
Popular Podcasts
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
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.
The Clay Travis and Buck Sexton Show
The Clay Travis and Buck Sexton Show. Clay Travis and Buck Sexton tackle the biggest stories in news, politics and current events with intelligence and humor. From the border crisis, to the madness of cancel culture and far-left missteps, Clay and Buck guide listeners through the latest headlines and hot topics with fun and entertaining conversations and opinions.