Skip to main content
MANIFOLD
[DEBATE] Can LLMs lead to AGI in <12 years? {2000 Mana bounty, potentially more}
6
Ṁ110Ṁ163
resolved Mar 27
Resolved
N/A

I aim to resolve this market as YES if good arguments are presented, regardless of my position in the market (in fact, i go out of my way to lose mana as evident by my profit graphs just to get some orthogonal value out of the market)

I believe that alignment is very important and short-term risks and long-term dangers of AI misalignment are p bad. However, I really struggle to internalize some of the vague arguments that people give when they talk about AGI <10 years and their refusal to explain how to overcome certain very concrete problems. And when asked to explain, the most common response i get is, Go read yudkowsky.

So, I'm left frustrated and confused at such conversations and the reading to learn more about it doesn't seem to be very formal.

My background is that i was a med student who switched to software and works at Google and have had hands on experience with several ML tools as well as familiar with the theory of machine learning high level.

i want people to make a case for short timelines, 12 years is pretty short for me. Why do you believe AGI can be achieved so soon?

I will try to debate, and point out specific problems. I intend to distribute the mana 1000 as 50/30/20 to the top 3 contributors who make solid cases regardless of whether I'm convinced or not, and as a gesture of genuinely wanting to hear good concrete arguments for why or how current technology can lead to general AI, here is a manalink of 50 mana for 20 people each so that people can claim just to participate in the market.

P.s. since this is a change my view kind of market please do not assume by default why a bad scenario is the default outcome (as recently done in eliezers market). Although i understand his argument for good aligned friendly ai being a minority in the space of possible agis, i also think that the methods we currently have to get to an agi by default exist within that tiny space and not just arbitrarily.

This market will resolve in a month, as it's purpose is debate and not forecasting, but incentives work better if conducted like this.

Mar 27, 10:07am: (DEBATE) Can AGI be achieved by 2035? (2000 Mana bounty) → [DEBATE] Can LLMs lead to AGI in <12 years? {2000 Mana bounty, potentially more}

Market context
Get
Ṁ1,000
to start trading!
Sort by:
Comment hidden