Excited to finally share this AI Reading List, compiling key resources on artificial intelligence & its long-term implications.
The list is divided into "80/20" resources and "deep dive" resources to help with suggested prioritization.
medium.com/@v_maini/ai-re…
Vishal Maini
1,688 posts
Joined December 2009
- MYTHOS FUND II IS CLOSED ❤️🔥
- o3 87.5% on ARC-AGI, but still only humans can ___... zooming out, you don't need to believe anything about existential risk to conclude that humanity's remaining pages in the book of history are numbered :(o3 @ 87.5% on ARC-AGI It was 16 hours at an increase rate of 3.5% an hour to "solved"
- Our new series is live! Machine Learning for Humans: A Beginner’s Guide to AI/ML🤖👶 cc @seriousssam medium.com/machine-learni…
- so uh.. sorry but i pranked my friend and he tweeted about it now 1 in every 40,000 living human beings has seen the tweet 🥲 turns out collective epistemics go out the window when a story is really funny be careful out there kids, trust no one
- Miles is a legend. One of the most thoughtful, original, and steadfast people I've met, and certainly the most arXiv-addicted person I'll ever know. I remember sitting at a conference with him, poolside around midnight, as he refreshed arxiv.org on his phone waitingI just sent this message to my colleagues, and elaborate on my decision and next steps in a blog post (see next tweet):
- “I honestly don’t think that there will be many AI companies in 10 years"
- As someone who spent 3 years on DeepMind comms, I have to agree here. It's one of the most risk-averse environments I've ever encountered. Let people say what they actually think (and what is actually true), and many good things will happen. Some bad things, too. It's worth it.Whoever’s doing comms for Gemini needs to apologize to Noam Shazeer and then resign DeepMind doesn’t get nearly as much credit and attention as they deserve, and that’s because comms is vastly underperforming capabilities You’ve got Demis, Noam, Jeff Dean, and Sergey Brin all
- since we've all agreed to retrofit DeepSeek as evidence for our pre-existing beliefs, i'll chime in funding the hyperscalers was always a subsidy, value capture was always going to happen at the app layer, this has barely started
- $UPST remains one of the most shorted (and IMO misunderstood) companies in the market. Why? My inefficient markets hypothesis here is simple: almost everyone stopped paying attention. I mean this quite literally. Remember the scene in Hunger Games when Katniss misses her first
- all the headlines re: AI safety mega-donors, longtermist PR campaigns, Elon tweets & the like lead to lots of outrage that AI risk is overhyped (or sometimes a false sense of security) in fact, insanely few people actually work on AI risk. ignore the noise, solve the problem
- $UPST is building the foundation model for lending. Nobody else is even trying.












