Who's Who of AI

Robert Hawkins

246 trust researcher @rdhawkins.bsky.social · 3,990 followers
NLP & language AI business AI research
Why they matter

Researcher with public evidence across NLP & language, AI business, AI research.

AI signals
4
past 30d
Sources
4
distinct domains
Discussions
6
past 30d
Latest signal
2d ago
View every signal from Robert Hawkins →
asst prof @Stanford linguistics | director of social interaction lab 🌱 | bluskies about computational cognitive science & language
What they're sharing

Articles & links

AI agents are checking the scientific literature and spotting decades-old errors www.nature.com/articles/d41...

AI agents are checking the scientific literature — and spotting decades-old errors nature.com
AI Weekly's analysis →
  • A Zhejiang Lab chemist's AI predicting boiling points clashed with a 75-year-old reference database; manual checks showed the database, not the model, was wrong.
  • The same AI spotted further mistakes in older papers and reference books, including a typo and incorrect values of century-old boiling-point measurements.
  • Researchers caution AI fact-checkers are not reliable on their own because the models make mistakes like humans do and still need manual oversight.
Read full analysis →
View on Bluesky · ♥ 11 ↻ 1 ↩ 0 · 6 from the directory shared this · 9d ago
↻ Robert Hawkins reposted
arxiv cs.CL @arxiv-cs-cl.bsky.social

Yingpeng Ma, Jianhao Yan, Bei Shi, Ka Hou Kam, Runnan Wang, Xuebo Liu, Yulong Chen, Yue Zhang, Derek F. Wong Can LLM Agents Stick to the Script? A Benchmark for Long-Horizon Consistency in Interactive Narratives https://arxiv.org/abs/2608.08160

Can LLM Agents Stick to the Script? A Benchmark for Long-Horizon Consistency in Interactive Narratives arxiv.org
AI Weekly's analysis →
  • NCP-Bench spans 100 narrative environments derived from movie synopses, with an automated checker that scores player agent vs. narrator agent consistency turn by turn.
  • The best model tested, GPT-5.2, maintains only a 42% survival rate after 20 turns of interaction under unconstrained user interventions.
  • Across six state-of-the-art LLMs, fact conflicts dominate failure modes with rates running from 40% to 68%.
Read full analysis →
 View on Bluesky →
↻ Robert Hawkins reposted
Ethan Mollick @emollick.bsky.social

Great experiment testing how good AIs are getting at very ambitious end-to-end coding tasks. Opus 4.7, in 14 hours, was able to build a software package that would take 2-17 weeks of human engineering. It cost $251. The models are still not perfect, but are improving fast. epo…

epoch.ai View on Bluesky →
↻ Robert Hawkins reposted
Ethan Mollick @emollick.bsky.social

A GPT-4 powered (& thus quite obsolete today) assistant for Pakistani judges increased the amount of cases they saw by 6% with no impact on quality. elliottash.com/papers/Mehmo...

elliottash.com View on Bluesky →
↻ Robert Hawkins reposted
Fintan Mallory @fintanmallory.com

If you're interested in socially applied philosophy of language, philosophy of science and/or machine learning, this might be interesting: dl.acm.org/doi/10.1145/...

dl.acm.org View on Bluesky →
↻ Robert Hawkins reposted
@dirkwulff.bsky.social

🚨 Out now in TICS How can language models help cognitive science? @ruimata.bsky.social & I outline 5 uses: mapping research fields, formalizing theories, cleaning up constructs & measures, predicting behavior across tasks, and capturing environmental variation. 🔗 doi.org/10.10…

Redirecting linkinghub.elsevier.com View on Bluesky →
Their network

In Robert Hawkins's orbit

Center = Robert Hawkins. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.