![Bodhi Profile](https://pbs.twimg.com/profile_images/1877986936419491840/GLlct_0O_x96.jpg)
Bodhi
@BodhiSterling
Followers
62
Following
761
Statuses
192
Blogs, Musings, AI. The truth will set you free.
Joined February 2014
@inductionheads Yeah Dario has made it pretty clear that he doesn’t think of COT models as a distinguisher. Anthropic is sitting on an insanely intelligent AI, the only reason they haven’t released it yet is because of “safety”
0
0
4
@RyanReeves_ Imagine if the value of stocks traded was booked as revenue. That’s how they book crypto
1
0
8
2025: ASI in Math & Coding but outputs are still mainly limited to a text box. Some agentic features are created but they still fail on tasks longer than a few hours. 2026: ASI in most digital domains. Context length is solved, agents can work on problems for weeks to months. Still need human feedback for judgement calls in poorly defined search spaces. 2027: Embodied Intelligence in humanoid robots starts to become commonplace. Recursive self improvement really takes off - AI is now better than all humans at all tasks.
0
0
1
@TidefallCapital There isn’t a grand conspiracy here. It makes sense that they’d get more efficient overtime and extend useful lives. The decrease is due to the pace that GPU’s are improving, so of course useful lives now go down
0
0
2
Literally can’t wait for Grok3 in my Tesla
@Scobleizer And next is Grok in our cars. I can’t freakin wait to bounce ideas around with it while that same computer chauffeurs me.
4
0
0
@01Core_Ben @eftegarie I love $KLAC, $LRCX, $AMAT, especially since fintwit hardly discusses them
1
0
1
$CSU.TO Bull Case: software gets extremely easy to build and maintain with AI, so margins go way up Bear Case: software gets extremely easy to build and maintain so companies can do it themselves and/or competition for vertical software goes through the roof. It's l likely that in the short term they benefit, I'm much less sure about the medium/long term
1
0
1