For AI leaders Doom is a form of hype

Did I see that I could get rich with bitcoins investing until 100k and selling everything? For sure not. But i already saw basic flaws of crypto when it came out.

Hyperloop? I cried out very early on how dangeriuos this is and how stupid and that no one wants to ride in a small tube without proper precautions (i was wrong here, some poeple really use that las vegas tube).

But lets be clear about LLM: The first thing which motivated me to really look into this was a random YouTube video of someone talking/writing to GPT-3 and ask it to simulate being a mysql server. And it simulated this flawless. A table creation statement, 3 inserts and a working select statement.

That showed me something: The chance that someone trained that 'thing' to be able to simulate a random mysql server was very low. I then tried it and the first thing not happened: I wasn't able to find the flaw. You know when Google IO or Apple Keynotes are happening and you get your hands on any demo and you live in a differen country or speak a different language or you are just not in San Francisco were most demos were maid for, it breaks fast. GPT-3 didn't break fast. I was able to interleave english with german words, no issue.

And i wasn't following LLMs before, i was following progress in machine learning and machine learning started to get more and more traction.

I can now prompt a tremendes amount of generic things to a neural network and it responds very convincing.

I have never seen technology advancing that fast and its quite clear what we have to still build around this new technology. We are not even were we will be and its obviouse.

Only when we build all the stuff (ai layer, ai platforms, ai chips, language and skill specific models, ...) and then starting to see any plateu we can stop keeping an eye on it.

And while we do that, everything just got a little bit better again and another interseting new thing was discoveed about LLMs, AI, AGI etc.

Right now we are in the middle of a reinforcement learning phase which will bring even more.

These 'no consciousness' thing already took jobs from people. I was planing to pay an blender artist money on fiverr for a project, since Opus 5.0 it was just able to do this for me without fiverr.

I don't think these unscrupulous guys are wrong. I'm following this area for a decade and we might have hit an inflection point: We have the compute, apparently we have the money as a society to build this massive amout of compute, we have also everything digital. It doesn't matter what it is, everything we have is digital. Every training material, every best practice tips and tricks everythign is digital.

You can be as dismissive about all of this as you want, but when the economy of a whole planet invest the most money it ever had into one area of research, you should try to keep an eye of where this is going.

And these random technologies? Random people invested in these for different reasons for sure, but the investment into AI is a lot broader with a lot more money behind it. The richest companies on the planet with the smartest people on the planet are doing one thing: Investing in AI.

Lack of technology alone will not wipe out the entirety of humanity, I agree. But we're not talking about just a lack of something. We're rapidly sprinting to give the fewest people the most power over the planet ever. I think it's likely to happen in the 21st century. What happens when the worst case scenario isn't just being left with nothing, but facing an entity that is so rich and powerful that it can use active measures to kill as many as possible? This is more power than any warlord or emperor could've dreamed of in the past.

> But as much as I think people like Elon Musk and Mark Zuckerberg are unhinged lunatics, I also think they're smart enough to keep people fed and entertained for pennies on their accumulating dollars.

So far, every time the billionaire class has been asked about how much money should go to them in society, their answer has always been a resounding "ALL OF IT". You don't get to that level of wealth while retaining a state of mind that allows you to toss a penny to the beggar. Reality is what's holding them back right now, but what would happen if their dreams of total automation eventually come true? How long from then until one of them snaps and decides to use state of the art automated military technology and weapons of mass destruction to wage war against the peasants (armed with sticks and stones) to solve the 'problem' once and for all?

To be clear, I'm not saying any of this is imminent or remotely likely to happen soon. But our society is moving in a direction where something like this could eventually be possible. Giving a human truly absolute power over the entire world has never been tried before.

It's about as much of a common thread as comparing Shohei Ohtani to a loan-shark's enforcer because they both swing bats at things.