Riggghht. It's so magical that even the ones that giant Fortune 500 companies are spending billions on are seeing zero return on. S&P 490 actually losing money. But the government has quantum AGI right? The scammers say "it's not circular roundtripping, it's vertical integration!!" God this snake oil is so pathetic.

---

All AI risk is with OpenAI aka in the private sector. High tech had fantastic businesses before LLMs, and if they disappeared tomorrow they would still print cash. The real risk on the public market side is for any small or mid cap that uses OpenAI’s API. Their API right now is stupidly cheap because VCs are picking up the tab. This won’t continue. Either OpenAI gets to AGI by ~2027, or they don’t, and they won’t. When they don’t, AGI hype dies, Sam Altman is out, they bring in someone whose sole job is to recoup investor money, & overnight their API pricing jumps 5x. Further this isn’t an Uber “the business model is there we just need to get to a certain scale” type of situation because sometime in 2024 training data size and model quality plateaued. And on the financial side time complexity and compute cost scale non-linearly aka models can’t get cheaper and get better. When you put all this together you see that the only ones at risk in a post subsidized-LLM world are small and mid cap companies that jumped on the hype train, and incorporated LLMs into their product.

---

You really don't need AGI though for this to be a transformative technology

---

It's hilarious that there are people out there so phenomenally retarded that they believe AGI is anywhere near. None of the current "AI" products are even capable of evolving towards AGI

---

They're betting on the law of large numbers. AGI absolutely has zero chance to be born in a lab where a couple of tens of thousands of experts work on it. But show the promise of AGI to a couple of billion people and eventually there'll be an autist that'll say: this sucks. I can do it better! And then go and do it.

---

Well let me ask you the same question as the other guy So you really think AGI is not achievable? surely it has to be achieved one day? Is there something you think would need to happen for it to be achieved? Or is there something that is stopping tech companies from achieving it currently?

---

So you really think AGI is not achievable? I know you said "at this point"....but surely it has to be achieved one day? Is there something you think would need to happen for it to be achieved? Or is there something that is stopping tech companies from achieving it currently?

---

AGI is science fiction at this point (much like colonies on mars and the Tesla roadster).  Current AI is a mashup algo that can only riff on stuff it's seen before. Real human intelligence doesn't need to parse the entire codex of human learning to understand things and it runs on a couple of Watts a day. Anyone who has had children (and actually brought them up) knows this. They're barking up the wrong tree. We're as far away from that as we were 5 years ago.

---

AGI is not happening with the current AI path we are down. And we have sort of plateaued on our current path.

---

Considering openai is pivoting into porn, agi is out of question of far away

---