You thought 2024 in AI was wild? Buckle up, yeah — 2025 is about to take ‘unbelievable’ to a whole new level.

If you’ve heard rumors about AI slowing down, forget it. Nothing could be further from the truth. This is why, and here’s how you prepare when everything changes forever:

People vastly underestimate how much energy and resources will ultimately pour into AI. This is no bubble — it’s not some speculative bet on a vague future.

Those who see that the holy grail, the answer to life, the universe, and everything, lies at the end of this road will stop at nothing. Because there is no price too high when it comes to unlocking our true potential.

But it’s not just about scaling computing power or building bigger models. It’s about fine-tuning and weaving together the capabilities we already have into agentic workflows.

How to Prepare for the Unprecedented

I won’t pretend to know exactly how the future will look, and I definitely don’t have all the answers. But I do know this: sooner or later, we’ll have a multimodal AI agent capable of thinking in text, audio, and visuals.

It’ll operate your devices more efficiently than you do right now. Most tools and software will become obsolete, making room for models that do it all.

My advice? Focus on the basics of psychology, language, and storytelling. Once we’re all using the same tools, what really sets us apart is how well we can instruct them.

And in a world where AI can fool you by acting human, the first thing you need to understand is yourself. The answers aren’t out there — they’re waiting within.

And it works like this: It’s easy to appear as a master as long as you know exactly how good you are. Being a musician, I’ve seen a lot of people play their favorite song, even though they’re not really good enough to perform it perfectly.

Meanwhile, a less skilled performer might choose an easier song but plays it 100% perfectly. That’s the person people will listen to. That’s the guy who gets the girl, that’s the dude who others will perceive as a competent musician, even though the first person I mentioned might be overall on a higher skill level.

Always Perform At 100%

If the song you choose to play is so complicated that you can only perform it at 90%, it won’t sound right, and you won’t look like a master.

But even if you can only play four notes and you play them to perfection, you are, in fact, a master at what you currently do. It’s as simple as that.

The same principle applies to graphic arts or anything in life, really. If you’re a designer attempting to create complex effects and only partially succeed, the output is garbage.

Always choose techniques and methods you can fully utilize, no matter how simple they are, and everything you produce will resonate with people.

The Fastest Way to Learn

Of course, to progress, you need to challenge yourself and experience some failure while honing your skills, but don’t overdo it. Aiming for the most difficult task doesn’t make you learn any faster; just the opposite, in fact.

In order to learn anything, your brain needs what I would call ‘news of difference.’ Let’s say your task is to throw a ball into a hoop. If you stand so far away that you only hit it 50% of the time, there is no way for your cognition to learn anything because it’s random, like the flip of a coin.

Likewise, standing so close that your success rate is at 100% provides no news of difference and thus, no learning. Research has shown that your brain learns best when challenged at around 75%.

You position yourself so far away from the hoop that 3 out of 4 attempts will succeed. Gradually, you move back while maintaining the 75% success rate. Now, your brain is lighting up with a lot of news of difference to work with.

But again, this is for actually improving your skill. If you perform or work for others, always aim for 100% success, because then you are, in fact, a master of what you do and will be perceived as one.

Of course, most people have no idea what hypnosis actually means. I’m not talking about some tricks on stage, and it doesn’t involve a swinging pendulum.

What Is Hypnosis?

Hypnosis is a meta way of looking at language. It’s not so much about what is being said, but about how and in what structure. By observing people who naturally excel at convincing others, we found patterns common to all of them, regardless of background and culture.

We extracted those patterns, and now you can apply them to whatever it is you want to accomplish, and it will work. This phenomenon has been studied for a while and is used in clinical hypnosis and hypnotherapy, endorsed by reputable medical organizations.

Why Is This Relevant to LLMs?

If I tell you not to think about blue, you will inevitably think about blue. The only way for me to prevent you from thinking about blue is to shift your attention towards red instead.

Large Language Models (LLMs), having learned everything they know from our language, behave exactly the same way. The more you tell them not to do something, the more likely it is that they will do it.

Experimenting with ChatGPT, Llama, Claude, and other high-tier models, it has become apparent that they not only react well to hypnotic language patterns but also understand them better than any of us.

Not because they read about them in articles, but because in order to understand language at all, they have to look at it from a meta perspective, and that’s what hypnosis is.

Level Up Your Prompt Game

You don’t need to pay for prompt engineering courses or search the internet for the latest prompts when you have the knowledge to construct them yourself. And luckily for you, there is a lot of material on the topic available for free right now.

Educate yourself, learn something new for a change. Watch a YouTube video on the basics of hypnosis, look up the works of people like Milton H. Erickson. Read a book on psychology or NLP, by which I don’t mean Natural Language Processing but Neuro-linguistic programming.

The more you do, the more you will understand how it can be applied to prompt engineering and how it is the last thing you ever need to learn in a world where AI carries out your instructions.

With recent advancements in AI and robotics, we will see humanoid robots roaming our streets and holding full-time jobs within the next 5 to 10 years.

Tesla, Figure, Boston Dynamics, Agility Robotics, Sanctuary AI, Unitree, and numerous other companies are in a race to develop the first humanoid robot capable of performing real-world tasks, with the goal of achieving mass production within the coming years.

Combined with the revolutionary progress in AI, particularly in visual capabilities, and initiatives like NVIDIA’s Project GR00T, these machines will be equipped to navigate the world and communicate with us, mirroring the scenarios depicted in the science fiction movies we grew up watching.

Allow me to explain: Our brain is designed to take shortcuts, the most efficient route from where I am right now to what I want. Throughout history, we have optimized for that.

Instead of hunting all day for a meal, you simply go to the supermarket today. No amount of reminiscing about the good old days can stop this optimizing process.

Artifacts of the Old World

For the most part, we have accepted this change, but here and there, we hold onto relics from the past that serve no purpose anymore.

Yesterday, I had to write a letter to a government institution:

Dear Department of Parks & Recreation,

bla bla formal and considered language bla bla

Sincerely,

The Dude

I ultimately decided to use ChatGPT to write this letter for me, as 90% of it consists of filler words and follows a format that is hundreds of years old, created during a time when someone on horseback had to transport my letter.

What once made sense is just baggage today. Since people have read and written these same words so many times, over and over again, we don’t have much patience for them anymore. And with good reason.

New Media Is Different

Many things in life are like that. Articles are still written with the physical limitations of paper in mind, and media content is still produced as if we only had a limited time slot, akin to TV or radio.

Much of that is unnecessary filler material that we are being flooded with now. So I say, embrace the short attention span, cut out all the baggage, and get straight to the point.

Of course, there is still room for long-form content. But don’t waste your audience’s time with things they have heard or seen a thousand times before. When we consume longer news formats today, we want raw, unedited, authentic content instead of spoon-fed and produced narratives.

I’ve been working in online marketing, SEO and branding for about 17 years. Consulting and managing presences with millions of followers. And while the media landscape is ever evolving, over the years the same patterns emerge again.

Here are the 3 types of influencers to look out for:

Famous in Real Life

People who are already famous attract followers and engagement regardless of their marketing expertise. Elon Musk once insinuated that because he has the largest account on Twitter, it automatically follows that he is the best social media expert as well. Personally, I doubt that.

The same holds true for people surrounding big accounts like that; they get pushed by proximity, and algorithmically, they are all anomalies. Schizoposting memes and hot takes on life won’t get your personal brand or business anywhere. Neither will the greatest advice of all: “Just be yourself.”

Great marketers can replicate this kind of success from scratch and for anyone because they have actually conducted A/B split and multivariate testing across multiple accounts, enabling them to achieve repeatable results. Everything else is just an opinion.

First and Lucky

There are other circumstances that make having a large profile not necessarily the best source for marketing advice.

People who were first in a particular niche, such as AI art, often rose to the top by default. They typically don’t know why their methods worked or why they might eventually become ineffective.

Being in the right place at the right time or getting mentioned by the media once can be advantageous, but it doesn’t necessarily make them great marketers either.

Hard Work and Longevity

For great advice on building a presence on social media, look out for profiles who have proven that their methods work, without falling into the two categories mentioned above. These are the people who truly understand how this works.

When I returned to Twitter a year ago, I only had a few hundred German followers left. Half of them hadn’t posted in years, while the other half unfollowed me after my topical pivot. My initial posts received zero likes. No one assisted me, I never went viral, and certainly wasn’t among the first. Yet, I managed to build a loyal following of over 5000 people, whom I consider to be the coolest people in my niche.

I don’t necessarily want to sell you anything here either. Just make you aware of how to assess if people know what they are talking about.

Of course, every rule has exceptions. There are people who understand how it works despite being anomalies, but a big number alone doesn’t automatically make it so.

This practice has been tested in New Zealand and the Philippines and will eventually roll out worldwide.

I personally don’t like it. In fact, if I wanted to come up with a method to harm X, this is exactly what I would do.

Combined with the other problems small accounts face these days, it doesn’t bode well for the future of this platform.

War on Bots and Human Casualties

I’ve been trying to be more positive towards X because I love what Elon is trying to achieve here, but I’ve never been a blind follower, and I can’t support these decisions. Just like the war on drugs, the war on bots will not work.

From high up on the mountain, this might seem like a good idea, but down here on Earth, I can’t think of a worse one.

Many people around the globe, including those in the EU, don’t use credit cards as much as Americans do, automatically excluding a large percentage of the world’s population from this platform.

Moreover, the worst offenders, operating the largest botnets, can easily afford this, leaving only the poor affected by it.

X Needs New Creators

Small and medium-sized creators are the lifeblood of every social network. However, on X, they struggle to grow, can’t earn money, and are unable to link to any other platforms for earnings. Now, with many people unable to join anymore, the situation is even more dire.

If there aren’t enough small creators growing to become big one day, this platform doesn’t have a future and will simply fizzle out over time. A new app will come along and sweep it all up.

I see a lot of people leaving already, frustrated with the way things are going here. And these are not people who categorically hate on Elon Musk; quite the opposite, they came here after the takeover, believing this to be the future.

But if new accounts can’t reach anyone, grow, or earn money here, it won’t work. Charging new users to participate will turn this into Elon Musk’s personal fan club. And I might even stick around for that, but this isn’t what the ‘everything app’ and the ‘town square for humanity’ should look like.

If you don’t know much about color composition, these are various tools and resources able to assist you. AI-based tools, in particular, are very useful in this regard.

Picking the Right Color

Certain colors are associated with different emotional states. If you’re not sure about that, you can ask an LLM like ChatGPT to list them for you.

But more important than finding the right color is picking one and sticking with it. This is all about recognition, so every touchpoint of your brand — your profile picture, your header, and your media — looks and feels like it’s coming from the same entity.

On websites like ‘Adobe Color’ you can browse and generate color pallets of colors going well together. Just like in music, these are not arbitrary and not really up to personal taste, they are governed by the laws of physics.

Use AI as Much as Possible

With these tools, you don’t really have to know the math behind it; simply apply what they suggest and level up your online presence significantly.

To distinguish yourself from other brands, be specific about your color choice; don’t use base colors such as blue. Instead, opt for a specific shade of blue, like royal, navy blue, teal, or cyan.

If you are not sure about their names, use ChatGPT again. It can list prominent color shades for you and also identify the shade of color you’re looking at.

You can either upload a picture containing a color and have it identify the color, or use a color picker, as found in various graphic tools. Simply input a color code such as #2b99ec, and it will tell you which shade it is.

This way, you can instruct an AI image generator in specific colors, always representing your brand.

Elon recently predicted the arrival of artificial general intelligence (AGI) for next year, and, of course, Gary Marcus won’t have it.

Gary initially bet $1 million, and Damion Hänkejh raised another $9 million to challenge Elon. But what do you think? Who is right?

Experts Are Divided

AI experts have been in disagreement on this point. Gary Marcus’s camp believes generative AI to be unreliable, and a new breakthrough is likely required to beat a human in any occupation.

Others, myself included, are convinced that all the breakthroughs have been made, and it’s just a matter of improving and assembling them, increasing the datasets, and scaling them to perfection.

I Believe the Singularity Is Near

Most people in Gary’s camp judge consumer-grade LLMs by the reliability of a single answer to a question. And sure enough, they make mistakes.

But as soon as you access them through the API, connect them to your own software, and utilize multiple agents checking each other, compartmentalizing tasks, and building recursive prompt chains, you know what’s going to happen. Because those systems don’t make any mistakes anymore and are on the verge of replacing entire industries.

I can’t say if a perfect multimodal AI system can be deployed by 2025, but I’m certain we are closer to it than a lot of experts predicted. AGI doesn’t have to mean the machine has to be as sentient and complex as we are; it just needs to do everything we do. Those are two very different things.

Ideally, generative AI extracts patterns from all of human knowledge, relying on our strengths. In doing this, it inevitably becomes better than any single one of us and soon.

The short answer is: Whatever you feel like!

Throughout our existence, our species has battled against depleting resources, striving to overcome material concerns. Now, we have a chance to break free from this endless cycle, though most people can’t imagine it.

Unlimited Energy and Resources

Think it through. Once we have an AI capable of performing nearly all human jobs out in the physical world, we can construct entirely automated industries.

The machine will gather the necessary resources and energy to sustain itself, maintain itself, make copies of itself, and perform any other tasks it has been instructed to do.

You end up with a process that doesn’t require us to put any work, money, or energy into it. It will simply create more at an exponential rate. And once resources run out, it will turn its attention towards the sky.

Imagine if SpaceX were 100% automated, capable of launching a new version of Starship every day until it works, granting us access to the entire solar system and all of its resources.

Distribution of Wealth

The only remaining question is: How do we distribute the gains of the AI revolution?

There is a flaw in our current economic system. Companies were supposed to pay taxes on all productivity. Our old laws could not envision entirely automated systems, so they coupled taxes to individual workers, rather than the work being done. That was a grave mistake.

Now companies are generating more and more productivity with fewer and fewer workers. What might be legal within the text of the law diverges from the spirit of the law, for the agreement was: all productivity is taxed.

Since that is not the case anymore, nation-states grow poorer while megacorps rise to unchallenged power. This process has been ongoing for quite some time; despite higher outputs, most people have less and less.

AI Reveals the Truth

But at least with AI, this is becoming obvious now. How is it right for companies like OpenAI and Midjourney to appropriate all of the world’s knowledge and then profit from it alone? Obviously, they have to give something back to all of us.

I don’t know exactly what the solution looks like, whether it’s some form of Universal Basic Income (UBI) or something we cannot conceive of yet. But our current way of doing things is over; it simply won’t work anymore, and it’s starting to leak through the cracks, as we are living in an illusion of a world long gone.

Purpose and Meaning

A lot of people say that without work, we will have no meaning, but I couldn’t disagree more. Waking up in the morning and having to think about hustling for food and shelter is the hallmark of an animal, driven by external forces.

But we have the potential to be so much more than that. We are here to create instead of simply reacting to the world. Automation has made us all work like machines in the last few decades, without purpose or meaningful impact. What people fear has already happened.

It is time to free ourselves from material concerns, so we can awaken to our vast potential, truly embrace our humanity, and explore what else is there.