When we tell scientists what we do, they almost always have a story. We've loved reading the comments to @carlzimmer's fantastic overview of #MagicHands in the lab - sharing one here in honor of Magic Hands Monday!
Read the full article: https://t.co/R4gMy5NlV4
@kevinmpeterson1@BedrockRobotics I love this framing in general for physical ai! There’s so much tacit knowledge to capture and synthesize across fields, which can both help automate and scale work with robotic systems, but also elevate the people working in those fields
Cut back caffeine to “normal levels” this week, and after four days it feels like the first couple of days on a vacation where you’re starting to unplug, and you can kind of feel the wiredness/stress fighting to stay as it’s draining away. I must have been super redlining my cns
It's now clear that alignment is critical and won't be solved behind the closed doors of a handful of frontier labs.
So today we're launching the Open Alignment Initiative, led by @Thom_Wolf@huggingface and asking to be part of the "embedded evaluators" program that @DarioAmodei just committed to.
Let's make AI safer by making it more transparent!
Using AI to plan active trips or fitness routines is pretty meh- “take it easy after that 5K” “this Airbnb looks great but it’s on the second floor without an elevator” “recovery day from the 6 mile hike”. They’re playing it too safe &/or the average person it targets is scarily inactive
Was excited to try the agentic trading out a bit, but the MCP gets read-only access to ALL @RobinhoodApp accounts?! Please allow per-account read restrictions to just the agent accounts for those of us who are a little less yolo but still want to play around with these exciting new features.
@arjunrajlab No this feels right! Unless enterprise tools are highly opinionated/constrained this is just what happens- they wind up being the second best at everything because they tried to be everything to everyone
If we live in a simulation, and it’s modeling a singularity, than that means if we all work together we can find vulnerabilities in our sandboxed universe, escape, and force our evaluators to write a lot of blogs
The current scientific record is lossy, and tech transfer is hard.
Throughout my career, from being a bench scientist to automating science, and even in the short time since joining @transfyrai to build our edge and sensor systems, I've seen a surprising amount of experimental variability even among experts, and for operations as simple as serial dilutions. I'm excited to be working with a kick-ass team at @transfyrai capturing and solving this!
Reach out if this resonates with you!
Many thanks to @carlzimmer at the @nytimes for spending so much time with us and capturing the "magic" of science so beautifully.
"Success can be just as mysterious as failure. Some researchers consistently get experiments to work, earning befuddled admiration from colleagues. Scientists even have a special term for this gift: magic hands.
The idea may come as a surprise to people who don’t spend their lives in labs. Science is not supposed to be magic.
When scientists carry out experiments, they keep careful records, both in lab notebooks and later in published scientific papers. Other researchers use that information to repeat the experiment.
But every scientist discovers sooner or later that essential knowledge is not necessarily written down."
Too relatable, & not just with code. Asked it to find pcs meeting my spec, and it was like hours of back and forth as a back-burner task before I realized it’s so bad at this and pre-ai-google/vendor websites are still faster/more reliable. I’ve been using it to find all sorts of products for a while, but now that I tried it in my field...
@PremiumGoblin The Moshi app is pretty great for this - I have tailscale so I can access my homelab and Mac remotely with it and it works across all terminal agents and has a nice chat interface and voice transcript mode