The AI crash is impossible - Change my View

[pablo_aka_god]

AI may never reach AGI.

but do you need AGI to drive a car autonomously?

do you need AGI to run a help desk for a bank?

to serve coffee? to run a cashier?

My answer is no.

May not replace all engineers, accountants and lawyers but marketplace could shrink by 50%.

The Silicon valley people loves to repeat as a parrot "ohhh but new jobs will be created like in the industrial revolution, we just don't know them yet"

For sure some new roles like robot service men may appear but enough to replace 60% of lost jobs or only 10%?

I want to be optimist about the future but i feel like I need shut off my reason to do it. Every time I use my reason can't thing anything else than "we are doomed"

Would love to hear more positive predictions! Don't want to be a doomer but my head takes my there when I look at the facts.

Edited 4 hours ago by pablo_aka_god

[Lyubov]

As I have done my own contemplation I've dimly had my own realization that AGI is an impossibility as long as scientists are asleep. It is not possible to create an AGI machine if you are spiritually asleep and do not understand intelligence. These guys now are like cavemen flapping their arms thinking the harder they flap they will fly like a bird. I could be wrong, I have more digging to do. 

[LordFall]

On 9/23/2026 at 5:56 AM, something_else said:

Yes. It is one of the things that defines us as a species. Most humans strive for it, some with more success than others.

I guess strive is a word I trapped myself with, I think you are wrong on this idea.

Most people do not work to build the assets necessary to build agency and ownership in the world. They pay their bills then distract themselves. They do not work towards owning cashflow businesses or other solid assets. At best they will match their employer's contribution to their 401k and a low percentage will ever learn proper investing. Financial ownership has a definition, it means to own assets and not solely liabilities. 

On 9/23/2026 at 5:56 AM, something_else said:

What is someone meant to do? "ChatGPT defend me and my family against the AI apocalypse" 

Someone can learn how AI agents work and use them each day to build value. Someone can customize open-source models for specific business workflows. It doesn’t take a master's in machine learning to do that; you can learn fine-tuning and agent integration from tutorials and documentation in a few hundred hours.

I myself considered before to get into the business of selling custom AI models to small and medium enterprises and will revisit this idea next year, it's an extremely lucrative and future proof niche. 

On 9/23/2026 at 5:56 AM, something_else said:

What is the pitfall you are talking about? It isn't clear to me from your comment, could you expand?

People do not provide equal value to the marketplace and most communists waste their time with intellectual narcissism instead of solving problems. I've gone to many communist discussions and even private meetings and I met a bunch of smart and dedicated people who waste their time complaining about society and cannot feed even 1 other person outside of themselves.

They'd rather talk about the evil bourgeoisie than start an organization that helps their neighbourhood, city or the world as a whole. 

On 9/23/2026 at 5:56 AM, something_else said:

That essentially leaves your argument at "it's a good thing if the rich people who just happen to control the AI continue to empower themselves and their dynasties at the expense of everybody else" which is pretty hard point to defend IMO.

I don't know if you realise just how different a post-scarcity world could be from our current one. It's possible money and wealth would cease to exist in their current forms. Our current tools barely work now, let alone in this theoretical world.

No I'm taking this conversation out of the realm of philosophy and into the real world. If you can snap your fingers and make it so everyone on earth owns AI equally tomorrow Friday September 25th 2026 that would probably be a good thing, please do that.

If you are a basic human like the rest of us I suggest you simply use the AI tools at our disposal to up-skill yourself for your future benefit and that of your loved ones just like I am doing and any other rational and motivated person I know is doing. 

Edited 56 minutes ago by LordFall

[something_else]

13 minutes ago, LordFall said:

Someone can learn how AI agents work and use them each day to build value. Someone can build their own propriety models. It doesn't take a AI masters to do that, you can teach yourself from Youtube in a few hundred hours. I myself considered before to get into the business of selling custom AI models to small and medium enterprises and will revisit this idea next year, it's an extremely lucrative and future proof niche. 

I think you could definitely sell that premise. A custom or retrained model you can realistically create is unlikely to be competitive with frontier models for most tasks though. Retraining or fine-tuning is somewhat old news now for most use cases.

You'd likely be better off running a context-engineering agency that configures the frontier models for companies through skills, plugins, custom agent loops, CLAUDE/AGENT.md files and other forms of external memory and context. Getting this setup right massively improves their performance on tasks for a business which can provide a lot of value. AI is all about getting the right context into the model just now.

On your more general point, I don't know how this protects you against the risks of post-scarcity AI-driven world. We're talking about a world where there are basically two classes of people, those with control over the means to produce value and those without. Being able to build custom models or tweak them does not put you in the former unless you earn so much money that you are competitive with the likes of the massive AI companies who actually control the most capable models.

14 minutes ago, LordFall said:

People do not provide equal value to the marketplace and most communists waste their time with intellectual narcissism instead of solving problems. I've gone to many communist discussions and even private meetings and I met a bunch of smart and dedicated people who waste their time complaining about society and cannot feed even 1 other person outside of themselves.

They'd rather talk about the evil bourgeoisie than start an organization that helps their neighbourhood, city or the world as a whole. 

Oh, right, yea on this point I agree. My point was not that everybody has equal value in the marketplace just now, my point is that everybody basically would in a post-scarcity AI-driven world.

17 minutes ago, LordFall said:

No I'm taking this conversation out of the realm of philosophy and into the real world. If you can snap your fingers and make it so everyone on earth owns AI equally tomorrow Friday September 25th 2026 that would probably be a good thing, please do that.

 

"bringing it into the real world" > "proposes highly unrealistic theoretical scenario" 

I'm only talking about a theoretical post-scarcity AI-driven world so much because that was the original topic of discussion. I have been trying to keep this discussion anchored to that original point. You said it was just my gut feeling that such a world is more likely to be dystopian than utopian, I think I have provided a fairly solid argument that it is very possible it will be, and that it is not just a gut feeling.

Quote

If you can snap your fingers and make it so everyone on earth owns AI equally tomorrow Friday September 25th 2026 that would probably be a good thing, please do that.

I'm interpreting everybody 'owning AI equally' to mean everyone has access to a frontier model that they have full control over with no guardrails, which cannot be taken away from them. If that's not what you meant then please correct me. But assuming it is, who knows if that would be a good thing. It's just your gut feeling that it would be good. You would then have dictators all over the world who now potentially have the capability to produce an extremely effective bio weapon, or to hack into nuclear launch sites and reactors of other countries. To assume that such a chaotic and rapid transformation of the world would be positive is, I think, quite unwise.

 

[yetineti]

You guys need to watch “Pantheon,” on Netflix.

[Joshe]

2 minutes ago, something_else said:

You'd likely be better off running a context-engineering agency that configures the frontier models for companies through skills, plugins, custom agent loops, CLAUDE/AGENT.md files and other forms of external memory and context. Getting this setup right massively improves their performance on tasks for a business which can provide a lot of value. AI is all about getting the right context into the model just now.

100%. It’s not easy at all to do this effectively when you’re dealing with a complex business/system. I spend a lot of time on this. We should set up a thread for this topic and share ideas.