sure but he put his money on training data for LLMs.
LLMS
-
Anthropic Introduces Automated AI Agent Code Review for Claude Code
By
–
🚨 Anthropic just dropped Code Review for Claude Code, and it just flipped the script on code reviews.
— Charly Wargnier (@DataChaz) 9 mars 2026
AI agents hunting bugs in parallel?
Yes please!
When a PR opens, Claude doesn't just scan the code.
It sends out multiple agents to find, cross-verify, and rank severities… pic.twitter.com/Yk2x4pCGS8Anthropic just dropped Code Review for Claude Code, and it just flipped the script on code reviews. AI agents hunting bugs in parallel? Yes please! When a PR opens, Claude doesn't just scan the code. It sends out multiple agents to find, cross-verify, and rank severities
-
Test Time Compute and Subagents: Optimizing AI Results
By
–
Roughly, the more tokens you throw at a coding problem, the better the result is. We call this test time compute. One way to make the result even better is to use separate context windows. This is what makes subagents work, and also why one agent can cause bugs and another
-
Introducing GPT 5.4 for Research Paper Analysis and Contextual Queries
By
–
Introducing GPT 5.4 for understanding research papers 🚀
— alphaXiv (@askalphaxiv) 9 mars 2026
Highlight any section of a paper to ask questions and “@” other papers for quick context, comparisons, and benchmark references pic.twitter.com/WmWJDoK5kfIntroducing GPT 5.4 for understanding research papers Highlight any section of a paper to ask questions and “@” other papers for quick context, comparisons, and benchmark references
-
Gary Marcus Criticizes Yann LeCun’s Glorification in AI Debate
By
–
my contempt originates in the kind of behavior I discussed here: https://
open.substack.com/pub/garymarcus
/p/the-false-glorification-of-yann-lecun?utm_campaign=post-expanded-share&utm_medium=web
… -
Gary Marcus Criticizes Yann LeCun’s Glorification in AI Discourse
By
–
see eg https://
open.substack.com/pub/garymarcus
/p/the-false-glorification-of-yann-lecun?utm_campaign=post-expanded-share&utm_medium=web
…, though it needs to be updated (again) -
Optimizing Claude Code usage for research with alphaXiv data
By
–
If you're using Claude Code for research: stop making it read directly from PDFs
— alphaXiv (@askalphaxiv) 9 mars 2026
We've introduced a SKILL.md that fetches structured, AI-friendly paper overviews from alphaXiv 👀 pic.twitter.com/AYg3n0hnB2If you're using Claude Code for research: stop making it read directly from PDFs We've introduced a SKILL.md that fetches structured, AI-friendly paper overviews from alphaXiv
-

ByteDance Seed Releases Helios: Real-Time Long Video Generation Model
By
–
"Helios: Real Real-Time Long Generation Model" More video gen techniques shared by ByteDance Seed! They show you can get minute-scale and temporally stable video generation in real time by training a 14B autoregressive diffusion model to expect and correct its own
-

AI and LLM Implementation: From Toy to Essential Tool
By
–
From a shiny new toy to an essential tool. Let’s discuss AI and the implementation of LLMs in the latest episode of Dev in the Details, featuring Dev Rishi and Ben Scharfstein 👉 go.rbrk.co/cewc0t
→ View original post on X — @predibase, 2026-03-09 17:54 UTC
-

SIPDO: AI that auto-discovers and fixes its own instruction errors
By
–
What if your AI could automatically hunt for its own mistakes and fix its instructions in a continuous loop? Researchers from UIUC, HKUST, USF, and http://
Starc.Institute introduce SIPDO to do exactly that. Instead of relying on a fixed set of data, SIPDO uses a feedback loop