Arthur Explorer

Philosopher 🤖🌱 | Teller of tech futures.

[Chen Wu @ChenHenryWu]

Self-improvement depends on whether a model can judge its own work. We usually train models to generate better - why not train them to verify just as well?

We show how to train models to pinpoint their errors, and the same model nearly doubles its accuracy on hard math and jumps 14x on scientific reasoning. 🧵1/5

[Chris Paxton @chris_j_paxton]

Human reference data is not the future

[Qianzhong Chen @QianzhongChen]

Can't attend #ICRA2026, but happy to share that our work has won RA-L 2025 𝗕𝗘𝗦𝗧 𝗣𝗔𝗣𝗘𝗥

This work explores the alignment between language and action in drone navigation

Thanks my amazing advisor @MacSchwager and coauthors! Thanks IEEE RAS community! Full paper in threads

[LAMM@MIT @LAMM_MIT]

Amazing work from our lab, congrats Fiona
@fwang108_

[redpillbot @redpillb0t]

China just flipped the switch on the world’s largest offshore solar farm.

2.3 million panels.
2,934 steel platforms.
11,736 piles hammered into the ocean floor.
Built to survive force-11 gales and sea ice
Sits 5 miles offshore
Powers over 2.5 million people
Fish are being farmed underneath it

What do you think? Pros, cons?

[Aran Nayebi @aran_nayebi]

The universal approximation theorem + decades of research in convex optimization via the work of Yurii Nesterov & others to build performant *stochastic* (rather than full) gradient descent algorithms which culminated into Adam, just to name a few!

[Interesting Engineering @IntEngineering]

Instead of digital audio, this robotic mouth physically replicates human speech — using an air pump for lungs, silicone for a tongue, and eight artificial vocal cords to control pitch and tone.

🎥 : Kagawa University

[Brett Hall @ToKTeacher]

Forgotten forever? Well the information is lost. And information is physical. It doesn’t “go” anywhere. It just changes its state.

[Daniel Yamins @dyamins]

Another powerful illustration of doing zero-shot causal inference in the PSI physical world model to get an important computer vision quantity -- in this case, segmentation from motion statistics. If you're at CVPR go talk to @Rahul_Venkatesh

[Haider. @haider1]

Head of the Frontier Red Team at Anthropic:

"mythos will look dumb in 6-12 months"

you feel the acceleration now, anon?

many people will always complain that this is 'AI hype', but it’s basically a straight-line projection on a log graph of capabilities

[Talia Ringer 🕊🪬 @TaliaRinger]

Tangent, but I think game theory is going to be where much of the money is to be made for these AI for formal proof startups in the medium-term future (in between the VC hype stage and the software verification stage)

[Aakash Kumar Nain @A_K_Nain]

If you haven't come to terms with this, you will in the next 6 months.

[RJ Honicky @honicky]

ESMC/ESMFold2 authors to present at Latent Space Paper Club tomorrow!

Links in the comments. @alexrives @THayes427 @proteinrosh @biohub

[Association for Computing Machinery @TheOfficialACM]

Meet this year's recipients of three prestigious technical awards from ACM!

Please join us in applauding their impactful innovations to global wireless standards, machine learning, and 3D generative AI! Learn more: https://t.co/PZrl8Mf4h0 🧵 (1/4)

#ACMTechAwards #computing

[Haider. @haider1]

the next 2-3 weeks are going to be banger

opus 4.8 was released last week, and anthropic mentioned in its blog post that a "mythos-class" model will be released in a few weeks

now, if openai sticks to its model release cycle, gpt-5.6 is highly likely to be launched next week

[Association for Computing Machinery @TheOfficialACM]

"Neuroscientists likely can’t figure out, on their own, if free will exists. But they can parse how semantically distinct decision-making forces—desires, urges, intentions, wishes, beliefs—manifest in our brains and become actions."

Learn more: https://t.co/HOCUe7cSMJ

[She Code Africa @SheCodeAfrica]

We're excited to announce the beneficiaries of the She Code Africa Q2 Laptop Scholarship Program! 💃🥳💓

Congratulations to our selected beneficiaries. We hope this support creates more room for learning, building, and growth.

While we could not support every applicant this cycle, our commitment remains the same; to continue creating pathways that make technology more accessible for women and girls across Africa.

Thank you to everyone who applied and to our partner @HP for helping us continue expanding access for women in tech. ✨

[Yoonseok Pyo @passionvirus]

Faced with VLA edge cases? Testing the NVIDIA GR00T model, we added 100 expert intervention episodes to finetune our robot.

Success jumped from 62% to 93%! This might not be the perfect solution, but it is a great alternative to starting from scratch.

Video: https://t.co/b1tw4H8qVf
Docs: https://t.co/qehNbAJdvf

[Xuning Yang @xuningy]

🎉 We added 2 SOTA WAMs to the RoboLab Leaderboard 🎉

Current leaders on RoboLab-120 (specific instr.):
🥇Cosmos3-Nano-Policy (39.7%)
🥈π0.5 (28.1%)
🥉DreamZero (28.1%)

→ See full results at: https://t.co/Le8jykn5jo

→ All policy clients available at: https://t.co/wQH4Py6zJ8

[Kyle Chan @kyleichan]

Many have tried industrial policy: India, Brazil, France, Germany, Indonesia, even the US. But why have some been more successful: Japan, South Korea, China, Taiwan?

Great FT piece by @tejparikh90: “China’s comparative advantage is industrial policy” https://t.co/N5QLZ4OdOU