Could GPT-4 recursively self-improve to AGI with the right cognitive architecture? [@Altimor, twitter]
2
70Ṁ1103023
9%
chance
1H
6H
1D
1W
1M
ALL
Altimor on twitter: Now 60% sure GPT-4 can be a seed AGI, with the right cognitive architecture. Defining seed AGI as a a good enough machine learning engineer / semiconductor designer to recursively self improve
(ht zvi)
Resolves if that happens, or when we get AGI and understand it well enough to claim it's impossible.
This question is managed and resolved by Manifold.
Get
1,000 to start trading!
Sort by:
I have no idea how he came to this conclusion, it doesn't seem particularly plausible.
He's replying to his post "We just had the most mind blowing breakthrough at Lindy. The formative insight of the company, I think. Couldn't be more excited, actually losing sleep over it.".
People are also trading
Related questions
Will GPT-5 be capable of recursive self-improvement?
16% chance
If GPT-5 can do recursive self-improvement, will it first be via fine-tuning on its outputs?
37% chance
Will LLMs such as GPT-4 be seen as at most just a part of the solution to AGI? (Gary Marcus GPT-4 prediction #7)
91% chance
Will OpenAI say GPT-5 is AGI?
10% chance
Will xAI develop a more capable LLM than GPT-5 by 2026
51% chance
GPT-Zero: By 2030, will anyone develop an AI with a massive GPT-like knowledge base that it taught itself?
33% chance
Will OpenAI's autonomous agent be based on GPT-4?
19% chance
Will GPT-5 have "the ability ... to autonomously replicate and acquire resources" per an ARC-like eval?
Is GPT4 sentient?
9% chance
Will GPT-5 make Manifold think very near-term AGI is more likely?
66% chance