|
|
|
|
|
|
by nemothekid
307 days ago
|
|
|
It's far too early to call anything with "AI" - and I will be unsurprised if we enter another AI Winter as LLMs turn into a dead end. If GPT-6, 7 and 8 continue to see diminishing returns on performance as we say from GPT-4.5 to GPT-5, then I see it becoming a winner take all of OpenAI. What I think is ignored is how at the time, all the investors were singing Meta's doom due to Apple retiring IDFA and META's stock was crashing. Then it comes out that Meta had spent that last year buying a ton of GPUs just before the GPU supply crunch and the stock completely rebounds. Zuckerburg seems incapable of being caught with his pants down. All the recent talk of Zuckerberg reeks of recency bias. He could completely screw up AI and he'd still be head and shoulders above every other founder/CEO. |
|
|
I really disagree with this take, if LLMs stop progressing you still have major FAANG corporations (like Meta!) paying gobs of money to Anthropic to increase dev productivity.
If LLMs only help the average office worker increase their productivity by 10%, that is still a huge multi-billion dollar market waiting to be captured.
You make good points about people doubting Zuck and Meta before and them being spectacularly wrong, I'm certainly not willing to bet the house against him right now. I think if he had been able to buy Thinking Machines I would have thought it was very similar to Instagram or WhatsApp.
But he wasn't able to, so now i'm not sure it's a done deal that Meta won't be wasting many billions.