Bradley Fehler
@bradfehler
Followers
628
Following
3K
Media
386
Statuses
7K
I am probably making some butter right now while helping researchers do their best work @Prolific
London, World
Joined March 2008
HUMAINE dataset of 33k human evaluations across 33 AI models! https://t.co/HU5OE3mXUO
0
0
0
To be clear: the problem isn't Copilot. It's the "add AI to everything" playbook. It's led to inferior and overpriced AI products. $30/seat/month for a wrapper most teams could build in a weekend with MCP or Claude agents - customized to their actual workflows. Microsoft is
177
454
9K
I didn’t truly understand how to build strong AI agents… until one paper snapped everything into place. Not a tutorial. Not a YouTube demo. A single arXiv paper: “Fundamentals of Building Autonomous LLM Agents.” It finally made sense why most “agents” feel like chatbots with
34
169
1K
Look mom. I’m embedding myself (and @Prolific) in the AI community with the friends at @encord_team
0
1
1
Holy sh1t they verified the results 🤯
We’re coordinating with @poetiq_ai to verify their reported ARC-AGI Public Eval score Only results on the Semi-Private hold-out set count as official ARC-AGI scores Once the verification is complete, we’ll publish the result and supporting datapoints
64
102
2K
Interesting, I haven’t really thought of this to be honest - “Why AI Won't Show Up in GDP” from @karpathy
0
0
0
Leaderboard illusion poster session is a wrap. Huge congrats! 🔥 I also hear there are is a mega rare leaderboard card you can still pick up from @singhshiviii and Oliver Nan.
2
8
99
Why your GTM is a product too. Great insight! https://t.co/B7aSI3b3Xj via @lennysan
0
0
0
In my opinion, the biggest risk in AI is concentration of power, capabilities and wealth. Let's fight this with open-source!
39
47
353
NeurIPS received 21,575 paper submissions this year. Our Agentic Reviewer, released last week, just surpassed this in number of papers submitted and reviewed. It's clear agentic paper reviewing is here to stay and will be impactful!
Releasing a new "Agentic Reviewer" for research papers. I started coding this as a weekend project, and @jyx_su made it much better. I was inspired by a student who had a paper rejected 6 times over 3 years. Their feedback loop -- waiting ~6 months for feedback each time -- was
74
285
2K
Excited to share new works on LLMs, agents and AI for science at #NeurIPS this week!👇 Thanks to my awesome students + collaborators. Look forward to meeting old and new friends in San Diego! Let me know if you want to chat!
4
14
129
@vedikaja_in @hud_evals @RobFarlow @Jerr_Wu If you’re building RL envs, human-data pipelines, RLaaS or synthetic worlds, we’d love to talk. We invest pre-seed → Series A (earlier is better), with one goal: maximize helpfulness-to-check-size. vedika@weekend.fund cr4sean@gmail.com
3
1
7
Why Humans Are Still Powering AI - an excellent interview with @Phelimb on @MLStreetTalk
https://t.co/Pvpe5bw7GS cc @Prolific
0
3
10
Prolific enabled automated recruitment and payment of participants in synchronous human‑AI experiments, targeting social scientists and behavioural researchers. read the research here
0
0
0
Meta just did the unthinkable. They figured out how to train AI agents without rewards, human demos, or supervision and it actually works better than both. It’s called 'Early Experience', and it quietly kills the two biggest pain points in agent training: → Human
87
297
2K
Holy shit. MIT just built an AI that can rewrite its own code to get smarter 🤯 It’s called SEAL (Self-Adapting Language Models). Instead of humans fine-tuning it, SEAL reads new info, rewrites it in its own words, and runs gradient updates on itself literally performing
641
2K
12K
Some generous companies in Toronto are funding three lectures on AI safety by Owain Evans on Nov 10, 11, 12. Tickets are $10 and are available at https://t.co/rFwHJPOKsq
63
92
683
How to design studies and get quality data with Dr. Simon Jones 🎥 Join our Senior Research Consultant as he shares expert tips to optimize data quality, and runs through the latest platform features designed to make launching studies faster and easier than ever. Register 👇
2
2
4
solo papers from great AI researchers are always such a treat
This is insane. New AI model from Samsung, 10,000x smaller than DeepSeek and Gemini 2.5 Pro just beat them on ARC-AGI 1 and 2 Samsung’s Tiny Recursive Model (TRM) is about 10,000x smaller than typical LLMs yet smarter because it thinks recursively instead of just predicting
10
38
684
🌀New work: Era of Real-World Human Interaction 🌀 📝: https://t.co/qB1roKk6Ou - RL *directly* from User Conversations - Organic replies + long-term history are learning signal - Trained on WildChat, beats RLHF at *user* level -> the future for personal Super Intelligence? 🧵1/6
9
71
360