AI - AI Dynamics

Nature Article: Complete Automation of AI Research

By

–

31 March 2026 18h52

Nature research paper: Towards end-to-end automation of AI research
go.nature.com/4rSOGFK [Translated from EN to English]

→ View original post on X — @sakanaailabs, 2026-03-31 16:52 UTC

31 March 2026

Americans Use AI But Barely Trust It, Poll Finds

By

@therundownai

–

31 March 2026 18h50

Some wild numbers from Quinnipiac's new AI poll: 51% of Americans surveyed use AI for research. Only 21% trust it. 70% say AI will decrease jobs. But only 30% are concerned it will touch theirs. Only 5% believe AI is being developed by people who represent their interests.

→ View original post on X — @therundownai

31 March 2026

Anthropic Launches Enterprise AI Services, Partners with Wall Street Firms

By

@theahmadosman

–

31 March 2026 18h50

The # of DMCAs Anthropic is about to send is going to be crazzzy I personally will spin up 3 Hermes agents to rewrite it in Rust, Zig, and C

→ View original post on X — @theahmadosman

31 March 2026

Detecting and Preventing Distillation Attacks in AI Models

By

@miketamir

–

31 March 2026 18h47

Detecting and preventing distillation attacks \ Anthropic bit.ly/3ORbiJ7 #AI #MachineLearning #DeepLearning #LLMs #DataScience

→ View original post on X — @miketamir, 2026-03-31 16:47 UTC

31 March 2026

New arXiv Paper and Official Launch of ARC Prize

By

@aihighlight

–

31 March 2026 18h46

arXiv paper: arxiv.org/abs/2603.24621 ARC Prize official page: arcprize.org/blog/arc-agi-3-launch [Translated from EN to English]

→ View original post on X — @aihighlight, 2026-03-31 16:46 UTC

31 March 2026

Convolutional Networks History and Attribution in AI

By

@garymarcus

–

31 March 2026 18h45

I have read him (a) that’s not true (eg see long history of convnets that precedes him, and (b) he brings receipts. seeya

→ View original post on X — @garymarcus

31 March 2026

ARC-AGI-3: New Benchmark Reveals AI-Human Gap

By

@aihighlight

–

31 March 2026 18h44

François Chollet just dropped the toughest benchmark that made every frontier AI look lost. ARC-AGI-3. 135 game environments built from scratch by game designers. No instructions, no rules, no stated goal. The AI gets placed inside and has to work out what it is even trying to do. Untrained humans cleared all 135. Every major model landed below 1%. Humans: 100%. Gemini 3.1 Pro: 0.37%. GPT 5.4: 0.26%. Opus 4.6: 0.25%. Grok-4.20: 0.00%. The scoring is built to punish shortcuts. A human solves it in 10 moves, the AI uses 100, the AI gets 1%. Throwing more compute at it makes no difference. For context: ARC-AGI-1 is essentially a solved problem at this point. Gemini scores 98% on it. ARC-AGI-2 went from 3% to 77% in less than a year with labs pouring millions into it. ARC-AGI-3 made all of that progress feel small. Announced live at Y Combinator in a fireside between Chollet and Sam Altman. $2M in prizes on Kaggle. Every winning solution has to be open sourced. Scaling will not fix this. We are not close to AGI. (Find link in the comments) [Translated from EN to English]

→ View original post on X — @aihighlight, 2026-03-31 16:44 UTC

31 March 2026