“There is zero chance any of these investments are going to turn a profit.”
This isn’t accurate. There’s a zero chance ALL of them turn a profit, but there’s actually good chance that one or more of them will return a huge profit.
If I invest $100 in a hundred companies, and 99 of those companies fail, you may think I’m a terrible investor. If that 1 company returns $50,000 on my investment though, I’m actually a fucking genius.
The real issue and what those AI dependent companies are banking on is that they can capture a user base and when OpenAI starts to reach the end of the road with LLM improvements and moves to the extract phase it can buy these little companies to ingest their user base.
Everyone else in that space will be instantly fucked since they will now be competing directly with openai while paying their margin but that’s the bet they are making.
They’re banking on diffusion eliminating the hallucination problem, but it’s too onerous to run very, very large models with it yet. Auto regressive LLMs are a dead end. One that is far away. LLMs are not and we will continue to learn a lot about them as we continue to implement them. Anyone who thought we were at a dead end should use the new Gemini. It’s like a GPT 3.5 to GPT 4 level of improvement.
“There is zero chance any of these investments are going to turn a profit.”
This isn’t accurate. There’s a zero chance ALL of them turn a profit, but there’s actually good chance that one or more of them will return a huge profit.
If I invest $100 in a hundred companies, and 99 of those companies fail, you may think I’m a terrible investor. If that 1 company returns $50,000 on my investment though, I’m actually a fucking genius.
That’s how venture capital works.
LLMs are evolutionary dead ends. Very expensive ones at at.
Even the companies doing things where LLMs are actually good at, rely on companies like OpenAI so when OpenAI goes down so will they.
The real issue and what those AI dependent companies are banking on is that they can capture a user base and when OpenAI starts to reach the end of the road with LLM improvements and moves to the extract phase it can buy these little companies to ingest their user base.
Everyone else in that space will be instantly fucked since they will now be competing directly with openai while paying their margin but that’s the bet they are making.
They’re banking on diffusion eliminating the hallucination problem, but it’s too onerous to run very, very large models with it yet. Auto regressive LLMs are a dead end. One that is far away. LLMs are not and we will continue to learn a lot about them as we continue to implement them. Anyone who thought we were at a dead end should use the new Gemini. It’s like a GPT 3.5 to GPT 4 level of improvement.