I think this take misses a major point, that is the question of why returns on investment in physics are diminishing. In the past centuries it used to be the case that new physical discoveries led to better technologies which in turn enabled new discoveries and so on. This is

You know we're in a new era of research when you can take a sick day and still hit the same measurable output because agents are doing things. The unmeasurable part which is latent thinking might take a hit, but that's hard to quantify. The real bottleneck now is simply

I feel like startups are using the term “physical AI” so they avoid the fact that they have to SHIP ACTUAL ROBOTS to have an actual business.

AgiBot founder Peng Zhihai says 80% of industrial tasks that humans can do but robots can’t is due to tactile sensing. https://t.co/ODzy9sFeR4

busy fucking day in AI - heres everything that happened: - Qwen dropped 4 ai models that can fit on your phone/laptop + as intelligent as OpenAI OSS-140B (but 13X CHEAPER) - an 18yr old kid sold his AI calorie-tracking app for ~$80M to myfitnesspal - a dude turned his macbook

Whenever I ask people what all the unprecedented amounts of software they can write with AI, they are writing is doing with I get replies that remind me of the Oscar Wilde quote: "The bureaucracy is expanding to meet the needs of the expanding bureaucracy."

Accelerated Python, written in x86-64 asm, initial release: https://t.co/Se7v04kHww