Do We Have a Future With AI?

I was watching Andrej Karpathy’s video, and it made me want to say something about it.
This topic is the hottest today, and all I hear is doomsday fear‑mongering…
And I get it: with any new, unknown technology that has fundamentally disruptive capabilities, we don’t know what to expect, and, as usual, we assume the worst… Especially AI being not just any dumb tool, but potentially a second intelligence on the planet?! That moves our problem space from infancy level to parent responsibility, to put it mildly.

Will Autonomy Come Tomorrow? 🔮

AI products will become increasingly autonomous and will need less input from us. Still, I think full autonomy is more likely not to arrive as quickly as everyone claims. We may stay in a pre‑autonomous stage for over a decade - or at least 5 years. I could be wrong, but the topic is complex and poorly defined, so I doubt we’ll reach it as fast as we fear.

How AI Will Change Learning?

For better and for worse, LLMs (Large Language Models) will shape education - not only for the young, but for everyone. Look around: how much have you learned from LLMs in the last few years? That will only increase. We’ll inherit some negatives, but that’s a small price for giving conversational AI to people whose only access to the internet is social media. I hope AI becomes a better instructor - for millions of people, than some spam messages via WhatsApp. High‑quality education should remain the goal, and AI‑driven platforms can help fill the vacuum where none exist, though we still need to solve illiteracy, so kids can at least read.

How AI Already Shapes Our Brains?

We’ll also need to process information faster as every stage of creation accelerates. In my first year with ChatGPT, it felt like sitting in a writers’ room - with ten screenwriters pumping out pages. I had to read book‑length drafts, most of it garbage, and discard it. I was fatigued and overwhelmed at first, but greed and impatience taught me to skim diagonally and decide quickly. Two years later, I still tire sometimes, but far less than before, and I know my capacity will keep growing. Yes, sometimes we let AI think for us, but AI also makes us consume more information and make more decisions, ultimately thinking more too.

Fortune: AI might already be warping our brains ...

Get ready to learn a lot - how to use AI, new products, more diversity, and a faster pace for everything.

Facts About LLMs

Superpowers

• Encyclopedic knowledge and memory
• Works fast

Deficits

• Hallucinations
• Uneven intelligence
• Limited context window / small working memory
• Gullibility

How Big The Market Will Grow With AI?

Improvement in LLMs and AI‑integrated products is accelerating. Get ready for an explosion of unprecedented diversity. My humble estimate is a market 100 × larger. We haven’t yet realised how much headroom technology still has. In my opinion we’re in the stone age of IT, robotics, and automation, with countless areas that don’t yet exist even in our imagination.

How Will AI Empower the Next Wave of Creators?

I want to talk about the opportunities: the possibility of a more dynamic, diverse world where things that were once impossible become easily accessible.

Just today, I was discussing this with a university teacher. He showed me a video generated in Midjourney with only three prompts and said, “The latest version is amazing!” He was right. I typed a prompt and the system produced exactly what I wanted in seconds, at a quality equal to what the animation studio I once worked for produced with 50 people, months of effort, and a huge budget.

Latest Midjourney videos from creators >>>

I remember feeling so excited because I could finally create animated shorts without doing things the old way - managing 50 people, raising money, balancing budgets, negotiating constantly - while still trying to stay creative. For me, the reward was always the creative part: thinking about the story, the visuals, the final look. The management side was frustrating and exhausting, but it was the price I had to pay.

Many talented people simply couldn’t acquire those management skills or weren’t willing to pay that price, so they contributed only small pieces to larger projects. Now, they can be storytellers on their own. They can choose their own stories, their own endings, their own visuals. We’re seeing more creators who handle the art and the narrative but don’t have to handle the bureaucracy. The same is true for games and other products. “But, it’s made by AI, it’s not human!”. Yes, some things will be created with minimal input and regard to the result, and we’re being already flooded by this slop everywhere, but that’s an ancient story, humans always were capable to create junk, and some junk has its own charm, you couldn't survive your doom scrolling without memes, would you?

AI Slop: Last Week Tonight >>>

But I’m talking about high value results, stories, films, and games. My favorite games are indie games, with unique voices. Do you say that’s bad? I can’t wait for more surprises.

My Learnings

• To get high‑quality output - say, an article - you need to follow the same steps you would with a human content marketer: create a project doc with audience, goals, outline, keywords, and tone. Then, write each paragraph separately and iterate until you’re happy with the content. Afterwards, add titles and other SEO elements. Skipping these steps always leads to subpar results.
• The same applies to any task. This may change, but even the smartest LLMs still need context, and only you can provide it.
• Too many or conflicting instructions can derail AI and produce unusable output. Use precise instructions that move in the direction you need. After many experiments, I’ve developed an intuition for which words confuse it - similar to managing people, where clear communication and context matter (though, unlike humans, motivation isn’t a factor).

How Products Will Look Like in the AI Era?

Today, thanks to AI, we build tiny tools - sometimes for just two users - because they meet our specific needs. They’re not bloated; they’re super‑simple. They do one or two things, nothing more. They’ll never be generic products designed to serve millions; they’re designed to serve me and maybe two other people. That freedom is exhilarating. I no longer need to raise funds, hire talent, or manage large teams. I can create whatever I want, for myself or for someone else.

From experience I can say: yes, AI tools can build applications for people who know nothing about code - but we’re not quite there yet. For now you still need to understand the basics of the product you’re building if you want to deploy it. Without some engineering know-how, most people won’t even try.

At the same time, AI will help you create something within your current skill set. It all depends on your experience. Not everyone needs to build apps. Many people will write, compose music, paint, or tell stories. Just create and take advantage of the opportunity you have right now - it works, and it’s amazing. Naturally, the complexity of what we build will mirror our experience. Those who’ve learned to manage complex products and logistics will keep doing so. Others will still be able to build surprisingly sophisticated things - just within certain limits. Ultimately, everything rests on imagination and experience. Many things I once thought impossible became doable because I cared enough to learn.

Curiosity is the only gatekeeper. What knowledge and tools do you need to pick up? There’s plenty to learn - tools, concepts, processes - but you can do it, now more than ever. That prospect excites me and fuels my vision of the future.

I've learned that our present - and our past - were built by people with vision who moved mountains to make it happen. It was rarely just one person; many people with different visions contributed pieces of a giant, complex puzzle. The same will hold true for the future: visionaries will shape it.

Things I’ve Learned That Matter in This New AI‑Powered World

• Software is changing in a fundamental way (watch the Karpathy video for details).
• When LLMs go down, people get stuck and can’t work-this is a new reality.
• We will have closed‑source LLMs and open‑source LLMs.
• Tool use, multimodality, and integration into existing products are in their infancy; they will grow to shape an entirely new ecosystem, transforming the internet and possibly swallowing it whole.
• Right now, AI mostly exists as a cloud hosted service, but personal‑computing LLMs are expanding. It’s likely that more people will run LLMs locally once the hardware catches up (give it a year or two).
• Why the military hasn’t led LLM innovation: LLMs need a lot of human‑generated data and a huge user base to cover training and operating costs-expenses the military or government can’t easily justify. Plus, imagine learning that the military scraped all your data to train an AI. (Some governments, like China or Russia, might not hesitate.)
• Agentic autonomy will grow, and AI will get better at completing projects independently.
• Much software will become partially autonomous.
• Software will be built for two users-AI and human-requiring a redesign of creation processes and UX/UI.
• Start‑ups will disrupt incumbents with lean, ground‑up products, so every market is up for grabs.
• We must learn how to design and build products that give both AI and humans clear access to information, actions, progress, and control.
• Many current products will vanish, but even more new tools will appear. Ex: If AI helps a creator produce an animated short, there’s no longer a separation of labour; software that handles the pipeline from script to publishing becomes far more useful.
• The web will become more AI‑friendly because AI will cannibalise search. Honestly, in the last two years I’ve used Google only a few times; I mostly search with Perplexity or OpenAI GPT. The only search I still do on Google is Google Maps.

Brace Yourselves

I am not the kind of person that will close my eyes on the problems, and yes, every new change comes with risks, and looking back, many of those risks were never solved 100%. We simply get used to the drawbacks and let the unlucky pay the price. Each new generation suffers the compounded consequences, yet we also possess a magical ability to adapt to the world we’re born into. The good thing is that young generations always fight to break the status quo. This conversation is inevitable and necessary; otherwise, the problems will never be solved.

It is a sin not to mention AI turning on us...

When Will AI Models Blackmail You? 

 or look around and see how Governments are using AI to control people already for a decade I must say… 🥺

AI will be used for good, AI will be used for questionable aims, and plenty will be used for harm - especially by large organisations, governments, and the military. Examples of what’s coming are drone warfare in Russia-Ukraine or robot‑dog police units deployed in Singapore during Covid and in China today. Our creativity applies to bad uses too, so we’re staring at a more complex future in terms of safety and human rights.

I will add to this vinaigrette some of my ingredients, enjoy! 🥗

More individuals building their own software will create new problems, just as social media amplified our worst impulses. Software reflects our individual cultures, drives, interests, and biases, and we’ll need to manage an experience similar to social media - but on a far bigger, more complex scale. As AI increasingly gains the power to act on its own, those goals, biases, and ethical ambiguities will inevitably be reflected in its decisions as well.

Good luck to social activists and governments. 🍀

Making that the future we want, depends on our vision, not our fears. Fear should be a check system, not the driver; otherwise we’ll end up living in a future designed by those who want to use AI for harm.

Will AI Change Just My World?

Since childhood I’ve been into the world‑bending stories - they completely consumed me following a long list of sci-fi books of course.

The same happened with the internet, social media, and later mobile phones, I was fully submerged. These technologies changed our lives in both good and bad ways. I could talk for years about my own experiences - how travelling in Asia was different before phones, Google Translate, and Google Maps. Now, to think of it, in two years that would be not yesterday, but 20 years ago… 😅

Now, what wows me is AI-and robotics, and big scale automation, it’s been for the last 15 years. First it was a quiet, geeky fascination. In the pre-OpenAI era, I tried every AI I could find, from random chat bots to the thumb‑sized app Replika. They were fun but mostly useless and disappointing. ChatGPT was different… When OpenAI released ChatGPT. I’m not ashamed to admit, It filled my entire field of vision: I applied it everywhere, in places it should and shouldn’t be 🙂. I went through all the versions, torturing myself and the servers as much as I could. I was watching with Nikro hours of AI news and discussions, it was truly exciting. It was reshaping my thoughts, on the philosophical level, ethics, and practicality. I was never that thirsty, to have so many questions answered… We jumped on board right away, thinking every day where we can use it, integrate and replace it, we've learned a lot in the process… and it was infinitely fun!

Now we’re a few years past that initial release (the ChatGPT moment), and I think we’ve crossed the point of no return. For me-and for more people every day-AI is an indispensable tool. I can no longer imagine doing many things without it, and there’s no going back. At the same time, AI lets me accomplish far more than I ever could even imagine. As AI products grow more sophisticated and more deeply integrated, my reliance on them only increases. Now the only thing that matters for me is to feed my curiosity, learn fast and dare to dream big.

And what about you, what’s your AI story?