aichannel

Curious what the crypto-sphere thinks about those timelines and outcomes

Curious what the crypto-sphere thinks about those timelines and outcomes

https://ai-2027.com/

I listened to their podcast with Dwarkesh and interestingly he didn’t seem to take it too seriously

Their theory that AI would be used to speed up research (including AI research) seems to be farther away than they think

I wonder how much of this is a result of advancements in LLMs vs RL?

But I think that once AI agents can work together to solve new problems, it gets very crazy very fast. Just unclear when that is

yeah I agree on the path, but the time it takes is what I think really changes the impact  it has

i think the focus on
- ace coding
- ace research

is pretty illuminating though. Unless there are unknown scaling blockers, that seems like a clear path to a more extreme take-off

I think he thinks about AI all day every day and is surrounded by the people doing the work so I believe he has a better idea of where the tech is / will be in the next 6 months

aside but curious if/why you care more for dwarkesh opinion?