Using AI to Supercharge Your Brain with Elias Kouloures

Welcome back to the Autistic Techie podcast.

Today I am joined by the incredible Elias Kouloures.

He is a neurodiverse AI innovator, marketing polymath, data scientist, and lifelong tech enthusiast over 20 years of international business experience.

Having lived on four continents and worked on nine cities, he brings a unique blend of creativity and technical expertise to bridge the gap between cutting edge AI research and

practical business applications.

Elias has collaborated on approximately 300 bookings for around 150 clients across seven countries, leading teams up to 35 people in roles such as chief creative director and

interim chief AI officer.

Elias, welcome to the show.

Thank you for having me, Shea.

It's a pleasure for me to be here.

Of course, I've been really enamored by your background, by all the things you talk about with AI and neurodiversity.

So tell us a little bit about yourself.

Well, I actually was amazed how much you took already in your intro.

So yes, I'm half Greek, half Austrian with a bit Italian Turkish sprinkles born in Munich, Germany.

So I consider myself European.

And as you mentioned, had a lot of jobs in a lot of places, four continents, 18 relocations until I turned 40 and yeah, 10, 11 perm employments.

as you mentioned, the last 11 years as a freelancer, starting out as copywriter and creative director in marketing, advertising, however you want to call it, and

transitioning into AI more and more since the last two, two and a half years since ChatGPT came out.

And as you also mentioned, over the time and my squiggly life path, I collected a lot of domain expertises because, yeah, I'm not just autistic, but I also have ADHD.

yeah it was a long winding uphill roller coaster ride but now it's going increasingly smooth and I have so many benefits from having learned so much

You have covered a of ground.

You said you moved 18 times?

Yeah, I'm an immigrant parent's kid, so a Greek father with Italian Turkish sprinkles and Austrian mother.

He comes from Athens, but his parents from small Greek islands, different kind of islands, and my mom from a small Austrian town, but they met in Munich, Germany because, you know,

they were all both for work relocating.

My father was helping build the infrastructure there, like the public transport, and he was like a craftsman, I think you call it,

My mom was secretary and both wanted to go into the big cities and Munich was the biggest for them.

It was economic wonder years in Germany.

Everything was booming, it was nice and they just met.

they are so completely different.

My father is so lively and maybe looking back he might have had ADHD because I always thought he was just Mediterranean and lively because

you know like Italians you know they get but the Greeks are very similar and Yeah, my mom was always the dry one, but now I also think that she might be on the spectrum I think I

have the autism from her and the ADHD for my dad and Yeah, first my...

they broke up my dad kidnapped me got me to Greece Took my mom over a year to get me back So my first language was Greek mom got me back and while I was settling in into primary

school She married an American and then I got back at night

at by relocating to Saudi Arabia, Jeddah, and yeah, lived in the Middle East for one half years where I learned English.

So I started learning English at the age of eight.

That's why I kind of decently speak it.

yeah, in the back and then later it was just the jobs, know, just relocating for the jobs from one city to another and being a workaholic because I love solving problems and

creating stuff and working advertising agencies.

That was quite fulfilling.

You've been doing a lot of crazy incredible things and you mentioned ADHD, you mentioned autism.

When did you discover that you're a neurodivergent?

What did that process look like for you?

Yeah, I discovered it pretty late because I'm probably much older than you so I went through my life on the one hand thinking the hyperactivity is due to the Mediterranean

part from my dad as said before and but there was always something else you know and at the age of 34 I finally went because I left permanent employment and started to become a

freelancer and I figured like because I've read about ADHD and I figured like okay let's try

that pill and immediately the doctor confirmed it and from my first Ritalin that I had my productivity skyrocketed like fivefold, not five percent but fivefold and I could do in

one day what took me a week before it.

I was already a workaholic so it was amazing but there was still this thing I was all of my life I was accused to play a role and like what do mean that's me I mean what kind of

role is it like?

and at the age of 40 or shortly before my 40th birthday, it was 2017, I read an article about Greta Thunberg in the Spiegel, which is like the Time magazine in America, it's the

leading weekly newspaper in Germany.

And it was one of those articles, know, maybe two images and the rest was just eight or nine pages just of text, text, text.

And I started reading it and there was no mention in the beginning about her stuff, just...

Her upbringing, her raising, and then they went into deep about her personality.

She's like this, she's like this, and it freaked me out.

I multiple times, I switched to the title if they were talking about me.

No, it's about Greta Thunberg.

Okay, so it continued, and on the maybe seventh page, they mentioned, because she has Asperger's autism.

It's like, what's this Asperger's?

The only thing, the only connection I had to autism was Dustin Hoffman, Rain Man.

For me, okay, that's autism.

And I was so naive because my mom, working mom, because the parents split up early, so she was working.

I came home from school, I just watched shows.

So basically my whole education was 80s, 90s American TV shows, like A-Team and whatever.

And I have never any, and also for me reading books back then was really hard because mostly boring, know, like just there are no pictures inside, you know.

And yeah, I mean I was even then of course I didn't immediately because it's also some people use it like it's cool now on social media Everybody a tick tock is nor a version.

So you so first of all, of course, I try to get an appointment but also the the system here in Germany is so Factor I wrote 80 different doctors to get a assessment no chance

Yeah

People wait years for an assessment.

That's not unique to Germany, unfortunately.

Yeah, but the thing is also, I assume that it's because I can't speak for all autistic people, but I assume that autistic people, least to my understanding, or Asperger's

Autists, that they like to research stuff.

So I researched for years and years and just did like all those tests and I always came out like 95, 96 percent.

And just looking at the traits that you have.

And also when I talk to friends of mine who were doctors or psychologists, you know, they always questions and you're much too hyper.

Yeah, have ADHD.

Plus I have autism.

So it's like being like a mixed race ethnicity, you know?

Nobody takes you seriously in none of your countries because you have this, so you don't quite belong there and have this, so don't belong there.

The same with autism and ADHD, you know?

autistic people are like, no, no, you're much too hyper for us.

And the ADHD people are like, yeah, but you're weird.

So yeah, it's like a third thing.

But yeah, after finding out,

and knowing it, know, first of all, I just am very open about it.

That's why I also talk with you.

And also I learned how to reap the benefits from it.

And that's turning out to be really great now with large language models in AI.

Yeah.

You mentioned AI a couple of times and AI has blown up in the last few years.

You mentioned ChatGPT earlier.

How do you think that AI plays into neurodiversity?

Do you think that neurodivergent people can benefit from AI?

Do you think that there might be some harms to it?

I'm curious to get your take on that.

Both, yes.

I think everyone, only neurodivergent people, but everyone can benefit from it.

And also, I believe everything has, I believe in being young, know.

Everything is up and downsides, yeah.

And for me, as I also encourage people later to go and use my free AI workshops that I have on my website, for me, I see it as especially large language models like ChatGPT

I see it as I specifically say they're like the most over motivated and smartest intern in the world who takes drugs and Has Alzheimer's, you know, so because you have this entity

that you can talk to I mean you can write you can talk to It always gives something back.

But if you give it like BS, it gives you BS back It's it's very knowledgeable about the world

It can basically comment on anything, but it also has like Hallucinations, that's why I say takes drugs ha ha and Alzheimer's because it has a certain Context window like it's

short-term memory.

So if you go over that it starts to forget stuff.

So and basically it's like a Logic-based a probabilistic partner you can have in your life and if you know how to communicate with it You can always get an assessment because I

think one of the challenges that especially

especially neurodiverse people have.

mean, just look, ADHD people are a lot in the world, like 8 to 10 % of the population has ADHD to some degree, and autistic people are about 2 2.2%, and Asperger's like 0.13.

Yeah, so for me, I always wondered, like, I mean, there's a lot of gas lighting going on, and I'm always wondering, is that normal what I'm doing?

Is that normal what other

are doing to me.

So first of all you can always have like an assessment, it's like or for example also use it to assess the reactions and actions of other people.

For example my ex-girlfriend, I mean I was with a very emotional artist, also different nationality together and I managed to get like a kind of a psychological evaluation or

like hints from the large language models out of it, but you can use it for everything you know.

You can use it as a thinking partner.

It's like imagine you have a very knowledgeable human 24 7 and the eyes are all kind of different, know, for example Elon Musk rock is a little bit cheeky and goes off the rails

while ChatGPT is a bit more, you know a bit more Let's say LGBT.

Yeah tempered LGBTQ friendly and and diverse friendly and a bit more leftist So yeah, you have someone that can guide you

that can give you assessments and also for example with ADHD it helps to structure stuff because my brain constantly jumps around and yeah so I encourage everyone, just autistic

people, because it has benefits and if you understand also that's why I use a metaphor if you understand the downsides then you also know that you always have to double check the

output and have to not take everything and copy and paste it into your I don't know homework or whatever.

Yep.

Something you mentioned I want to emphasize for people who are listening.

You mentioned not just typing to ChatGPT or Claude or Anthropic or whatever it is, but also talking to it.

Can you talk a little bit more about how people can talk to AI?

Because I think people are mostly used to text-based and web-based interfaces for doing this.

But what are some other ways in which people could leverage AI, like talking to it, or other maybe nontraditional inputs?

So it's becoming more and more ways to do it.

In the beginning, it all started out with text, with written text, because it was the easiest.

So you could type in a bit like a Google search and you'd get something back.

But now the AIs, the Dutch language models, are getting multimodal.

This means they understand images, they understand video, they understand sound, because what actually happens in the background, those neural networks are mathematical

multiplications.

take something and they cut it up and then give it a mathematical representation and then they can use it.

it started out with text and now most, I mean it depends, ChatGPT you can talk to, you can use the app, can talk to also Google's Gemini, Anthropic, I think not yet, but I there's

so many and most are even free.

So you can try it out.

And I said they also have like, a bit like a human, they different kind of personalities,

depends on the data they're trained on and on the guardrails they have and the system prompts.

example, Anthropic is the most safe.

And so you can use it by typing.

You can talk to it like I'm talking to you now.

But also use it, for example, to try out job interviews.

can film yourself and you can also tell it before, just with a very short prompt with a few lines.

You can say, I'm like this kind of job and I'm applying now to this kind of company and ask me questions.

And then you can actually go through and also have it give you feedback to it before you talk to the HR person or to the tech person or whoever you talk to.

So in a said.

basically anything that you any any situation any challenge any task that you can describe with words with written words or with with audio with spoken words you can use it for

yeah, so It's I think easier to say what you can't use it for and also the use cases are getting more and more so any and that's my suggestion or my recommendation is to Break

down whatever job you have breakdown

into small manageable tasks that you can describe and then over time you learn what kind of tasks you can give to the AI and whatnot and part of it you need to just experiment

because it's called the jaded border of competency because some things that humans can do AI struggles with and other things AI is extremely good and much better than humans

already so for example you can you can use

use it to decipher like handwriting.

I don't know, you've got a doctor's note, you can't decipher.

What does the doctor prescribe me?

Just take a screenshot and give it to AI.

Or you have a fridge and you don't know what to cook.

Take a picture of your open fridge and say, oh, what can I cook with it?

Make a few suggestions, you know?

So basically everything that you can give it in any kind of form that you can digest, text, video.

And now they're also starting to tokenize smell.

So soon it can also smell.

a new one.

How does it input that?

How do you do that?

So I only can tell you that there's a company that started to do that, know, to basically tokenize it.

And I said everything is mathematics in the end.

The AIs are just metrics multiplications and the neural networks are kind of built like the human brain.

we, like a human brain, you know, I could cut up your brain and I would see, okay, there's this flashy meaty thing.

And if I go with the microscope, I could also

see all your neurons, but I couldn't tell which of your neurons does what.

mean, now, course, science knows, okay, the thinking part is in this area and the limbic system is more for this and this.

Yeah.

But the AI is pretty much, it's like a field that is still being, is very new and it's being explored, but we, it trains itself.

You you just shove in data for weeks or months and it learns it itself and it adjusts the neural networks and the weights.

itself and where exactly what information is.

It's not possible to know because they have billions of parameter.

So we have billions of points that have a mathematical number and then you look at it.

how, I mean, we're going to find out at some point, but the whole large language model era started basically in 2017 with the invention of the transformer architecture.

So it's basically pretty, pretty new and

It's just amazing to see the way we've gone and the progress we've made.

That's pretty crazy and incredible stuff.

I use lot of AI for Dungeons and Dragons ideas, for puzzles, sometimes for illustrations, for combat scenarios.

I have actually found it as a really helpful coach for for D&D.

And had I had access to ChatGPT-like platforms when I was at my first job, I think things would have gone a lot differently for me as far as me being able to talk to someone like a

career coach like that.

I have worked with a career coach in the past, but...

I think AI, you set it really well, like 24-7.

That would have been incredible.

How have you used AI on your work?

And I obviously know that you can't talk about every little thing that you've done.

Some of that stuff is under wraps or NDA, but what are the ways in which AI has helped you put out really awesome work?

So AI has helped me in so many ways, I just need to structure my thoughts.

I use it pretty natively and for anything, for example, for brainstorming, for summarizations, for when you just mentioned D&D, know, what I would do if I would use it

for D&D, I would just think, okay, depending what you want to use it for to train or make a more, because AI is very good at recombinatory creativity.

Like, for example, write an email like Shakespeare.

Yeah, but if we take this to D&D you'd also say okay create us create for me a D&D Adventure that plays in the world of Middle-earth, but written by Quentin Tarantino.

Yeah, or or Pick whatever or think like okay, like I'm an evil Dumbledore from from from Harry Potter I know you just you have to give it something and then it can combine it in

amazing ways and That's what I use it for.

No, I

I use it to because as an autistic person and I assume other autistic person people are similar that I love detail, you know, and and that's also one of the strengths that we

have because the more context and detail you give it the better your output is because It is trained even though it's trained on the whole internet It is just a condensed

compressed version of it.

So for example, just let's say

I don't know Sam Altman the boss of open AI.

He's a very famous person or Donald Trump just to name someone yeah, so so so there's a lot of information about Donald Trump and It's it knows it yeah, but even though it's not

everything is in there But because he's mentioned so often it knows him by heart, but if you ask ChatGPT about me as a person without internet access it most probably won't know

me because In the end after the training

session after after 80 or 100 terabytes of being trained for weeks and weeks and months the file the neural network file that comes out is is is a thousandth or a ten thousandth

time smaller so it's like an image imagine image you have an image of the whole knowledge of the internet and then you compress it like you have a huge image so so that's why

Yeah, I always feed it the information I want that it uses.

Yeah, and I always do it, for example, when I work with a company.

First, I do with Perplexity, a very detailed research.

Yeah, and sometimes I think my record was that Perplexity looked up almost 200 sources and then condensed it.

And then I have the information about the current status, which is usually like a McKinsey level analysis that McKinsey back in the day is charged like two or three weeks of work.

And I have that in five minutes.

Then I use this data for example What I did for a German telecom which are telecommunications provider.

This was just a fun project I did out of boredom I just did that for them and then I did the same for a deso cup Gemini and Accenture and like two other three four companies and I

had their data sets and then I put it together and Said AI.

Okay.

Now, please make blue ocean strategies, which are basically new business models where

Where there are no competitors and I also gave it like the highly detailed analysis of all those things It's like I gave it like 60 pages of analysis and said like use that and make

new business models for telecom That makes it hard for those other companies where you also have all the information to compete Yeah, and even on my website the thing and I do

this just for fun just to experiment I constantly six to seven days a week I mean I've been now I didn't take a day off the last four weeks because I

constantly you know maybe I might be addicted to the productivity that I have but or you know yeah

Yeah, or another thing, like a tip also for the listeners is what is really good.

I mean, you can do it for brands.

you could tell it first analyze how does Nike or IBM, how do they talk?

Yeah.

And then you can write a text and give it the adjectives, you know, of the analysis.

And then you can get the output in their way, but you can also use it for practical things.

You know, for example, if you want to write a letter to your landlord, you know, that sounds sophisticated and professional,

your job application.

You can just tell it exactly who you want to address because for me with my ADHD you can tell I talk too much yeah and and you can also filter it down to the essentials you know

and that's what I mean with a partner because what people underestimate for me you know is the AI has learned all of the internet and of course there's a lot of BS on it but also

many books about psychology and people

And if you want to address somebody, a person, or you can even make a marketing persona that buys a certain product, know, the AI understands the psychology of this person and

can write targeted for this person.

And if you apply for a job, you know, at a tech company or at a, I don't know, at a law firm, it will write a different application.

And you can also tailor it by first analyzing what are the

pain points of this company and then look into my CV and write exactly how I can help them alleviate, mitigate those pain points, you know.

So, and that's what I mean.

That's why you need to experiment to understand what it can do and where it fails, you know.

And I said recombinatory creativity, good, where it's bad at the moment is creating something that's completely new.

For example, you know what a Dyson sphere is?

Yes.

Yeah, so I mean just for the listeners Dyson's fear is a theoretical thing that Advanced civilization might build around a Sun to harness the the Sun's raised the energy Yeah, so

this is just a theoretical concepts that astrophysicists came up with but we've never seen one so far Yeah, so the only artist drawings exist because this doesn't exist and if you

ask like mid journey or Leonardo or whatever diffusion model use to draw it

they have problems with it and it doesn't look really good because it doesn't exist.

And that's what I mean.

If it's a recombination of certain things, it's really, really good and you can get, like a monkey that goes skiing, yeah, okay, it knows what skis are, knows what a monkey is and

then there will be a nice image of it.

And as I said, just use it to...

free yourself up to have more time for creativity, for the really genuine stuff and the original stuff and use it for the repetitive stuff and also the stuff that is

recombinatory.

And if you break your whole life, your whole work and your private life down to the small tasks, then after a while, it's just like riding a bike or driving a car.

You just don't think about it, you just use it.

And it just makes fun.

It's a joy because I just love solving problems and getting stuff done.

That was a really thorough and detailed answer.

And I think that speaks a lot to the level of AI that's happening.

Like, that's not a straightforward answer.

And I don't think I was expecting to hear like a five minute answer to the question, but I think a five minute answer to the question was what we needed.

And don't worry about talking too much either.

And don't feel like because of your ADHD you have to talk less.

Like this is a subject where you're very passionate about it.

You're very knowledgeable about it.

So who am I to say to talk less?

Like, it's just not...

what I think is necessary and I'm glad that you are so excited and eager and passionate about it because this is the world that we live in.

We live in a world of AI and we do live in a world where these tools exist.

And to be honest with you, like we can't ignore them forever.

I think you can be thoughtful and careful about how you leverage these tools and some people may choose not to let it take over their lives, but I think at some point or

another,

Generative AI, LLMs, are going to come up in your life, whether you want them to or not.

And the sooner that folks have an understanding of how to meaningfully use them, the better off they're going to be and the more prepared they're going to be when push comes

to shove.

Yeah, true.

And just to round up for me, I see it as a tool.

me, I see AI just like a car or a bike or a toothbrush.

It's a tool that has up and downsides.

And I use it to do stuff quicker than before.

But it also empowers me to do things that couldn't do before.

So basically now I can do not only copyright and creative direct, I can do the whole spectrum of a traditional advertising agency plus a consultancy like McKinsey plus a film

production company.

in one person in record time and and and and that's something that also I think makes you very valuable you know also the listeners I said all the many listeners probably already

have like daily issues with because they're different and and I get that so it helps with that but it also makes you just so much more productive and not just for work also in life

you know because you have chores that you don't like and I for example I also

wrote like legal papers, you know, and got myself out of trouble many times.

And if you know how to prompt it, you can get pretty good results.

Important caveat that ChatGPT and things like that should not be used in place of a lawyer Please always consult professional legal advice if you if you need to I I have asked those

sorts of things questions before I have not asked is the draft up Legal wording for me as much as I want to I know that it can hallucinate things sometimes and I always would be

ever people ask me about how do I use this this and that for ChatGPT?

I'm like

Here's how you would use it, but like always like check with an expert or check with a professional before you kind of go off blindly and unfortunately I have had people who do

go off blindly and it hasn't always worked because it's still not completely accurate or bulletproof.

Yeah, I mean, if I may give two tips, two pro tips.

So yes, it's right, of course, be sure legal representation is important, but as a shock and awe, if somebody tries to mess with you, then you can, I'm nice person, but I can also

be very mean if somebody tries to take advantage of me.

Because as an autistic person, I'm very naive and I get taken advantage of, and that's why now I'm at an age, if you try to mess with me, I get very angry.

The autistic rage.

So the first tip is when you get legal advice,

is use like the search function and ask it to search for the newest legal things.

You can do it in ChatchiBT, but also Perplexity.

And then if you get some kind of recommendation with the paragraphs and everything, then give that to a different large language model and ask it to double check it, you know?

So for example, get ChatchiBT and give it then to Claude and ask like, is it right?

And then you drastically reduce the failure rate and that's why if you have a good system,

that change those things together that is almost at a zero point hallucination rate.

You know if you use it properly and if you have like them checking up on each other.

That is interesting, like having a cross reference.

I've never considered that.

Now I'm gonna have to start doing that for a lot of questions that I'm asking it.

That's definitely a...

Obviously the internet of course, knowing how to search things on the web is a very important skill.

So don't let it be said that you can't still search stuff on the internet.

Elias, if people wanna find you, they wanna learn more about AI, they wanna catch up all the stuff you're doing, how can people find you on the web?

Well, I have a difficult name, but I think the best way is to go to my website eliaskouloures.com and on my website you'll find my link to my LinkedIn.

I'm probably gonna add my X or Twitter link there.

You can also use my name and look me up on YouTube, but basically the website also links to YouTube and stuff.

there's a lot of free material on it.

First of all, I recommend everyone who wants to learn about the topic just to go to my AI.

have keynotes, AI data science keynotes, because they're all for free.

You don't even need to give me your email, whatever, I just share it.

And also have a lot of GPTs that you can use, know, that help you.

For example, I have a GPT that creates in seconds advertising campaigns for whatever you have, know.

you could use it for your podcast, but if somebody is an artist, can use it.

yeah, so I use my knowledge from the creative industry and the tech knowledge.

and I just built a system that uses 300 award-winning campaigns as benchmarks and then mangles it up with your whatever project and creates a new campaign.

yeah.

I'm gonna have to start doing that now for my AI usage.

And I will also make sure that all the links you mentioned are in the show notes, so wherever you're listening to this, can find those links easily wherever they are.

Elias this has been an excellent episode.

I have learned so much about AI.

I hope our listeners did too.

Thank you so much for coming on.

This was an amazing episode.

Thank you for being here.

Thank you, Shea it was a pleasure.

one last tip for all autistic people, get the books from Robert Greene, like The Laws of Human Nature.

That's like, I mean, I learned most of the chapters or most of the laws.

I the hard way during my life, but if you're a young autistic person, Robert Greene, The Laws of Human Nature will tell you how humans really are.

It's on my to-do list to read and I'm definitely gonna check them out.

Thanks again.

Thank you.

Have a good day.