Will Knight

@ErenChenAI amazing!

If robots legs are too short to climb the stairs, try this 😂

@ShayneRedford @alex_pentland @sarahookr @PeterHndrsn congrats!

Defended my PhD last week! 🎉🎉 Thank you to @alex_pentland @sarahookr @PeterHndrsn and everyone who was a part of it! Recording here: youtube.com/watch?v=A7BiI1…

@S_OhEigeartaigh Fair points. I am concerned that we really don't know how psychologically harmful it can be in general. Like other people writing about AI, I am inundated with worrying messages from people who seem to be experiencing mental health issues.

I do think there is a case for special caution around u18s given less experience, less time to develop critical evaluation skills, more susceptible to influence, still mentally and psychologically developing to a greater extent than adults. I just think presenting this book in this way isn't the greatest way to support it, and jumping from that to such a strong policy on AI (when the book isn't even about AI) also a leap.

Worth knowing Haidt's claims in this book are highly contested and criticised by experts in the field, for overstating the strength of evidence, presenting correlation as causation, etc. (Though certainly there are causes for caution around teens and social media, and u18s and AI where there's a lot we don't know.) Always feels like there's a weird imbalance with Anthropic. Parts of their policy/social impact portfolio and positioning are meticulously well-researched and carefully couched (e.g. see the economic stuff, and of course their safety/red-teaming work). Other parts seem to reflect strong personal views by leadership with far less care and scrutiny having been applied. It's ok for leadership personal views to influence company posture. But where those views are affecting activities of a company as central to the field as Anthropic, and as influential on national and international policy as they are & will be, it's important that these positions and the evidence/arguments underpinning them to be subject to robust internal and external debate and scrutiny.

The new White House policy requiring green card applicants to apply from outside the US is a capricious attack on legal immigration. It will hurt families, leave us with fewer doctors, teachers and scientists, and hurt American competitiveness in AI.

@hlntnr @SusanShirk1 Great point. The damage being done to American science right now is so much more important than reducing regs for rich companies.

A lot of AI/tech execs have been vocal about the importance of competing with China recently. Seeing who speaks out about this new green card policy will tell us a lot about who *actually* cares about US competitiveness, and who just uses China as a pretext to oppose regulation

The recent NYT opinion piece by @tab_delete is so fascinating, beyond just the way AI is upending education. "What really is the distinction between a nameless, faceless human you’ll never meet except over the internet and a statistical approximation of the same thing?". nytimes.com/2026/05/17/opi…

@KuphDev @Ken_Goldberg @spencerhuang_ nice!

@willknight @Ken_Goldberg @spencerhuang_ Indeed. I feel like I’ve got superpowers with my own robotics livestreams with the SO-101

It is truly mind-blowing how capable Claude and Codex now are at configuring and controlling robots. My own experiments (with a lerobot 101) point to a fascinating new trend in robotics that could accelerate adoption. Many thanks to @Ken_Goldberg and @spencerhuang_ for their insights! wired.com/story/i-gave-m…

@mkemka_ @Ken_Goldberg @spencerhuang_ Very cool. Seeed studio is amazing.

@willknight @Ken_Goldberg @spencerhuang_ Still trying to test it but it is quite powerful

impressive, but really looks like child labor.

@mkemka_ @Ken_Goldberg @spencerhuang_ It looks amazing. I will have to save up for one!

@willknight @Ken_Goldberg @spencerhuang_ Tried the b601?

Thx @willknight! Agentic Coding for Robotics: an exciting project w/: @letian_fu, @uynitsuj, @k_elrefai, @ethankou24, @HaoruXue, @_wenlixiao, @RavenHuang4, @guanzhi_wang, @drfeifei, @GuanyaShi, @jiajunwu_cs, SSastry, @yukez, @DrJimFan, @spenserhuang capgym.github.io

whoa!

@ClareWilsonMed 100%

@willknight Is that how you usually exit the dance floor too?

woah I think someone grabbed a video of me from my cousin's wedding to train this Unitree G1.

very curious malfunction here.

This doesn't define human intelligence, in fact a human's ability to know when it is wrong to manipulate someone might be considered a pretty important element of social intelligence. But anyway, ever since ChatGPT LLMs have essentially been optimized for persuasion...