stechert

Wrote search @Splunk, planner/schedulers/vision/datamining @NASAJPL, robotics @UCLA. Opinions mine.

[Taelin @VictorTaelin]

@midjourney the bravest pivot in history

[stechert @stechert]

@paulg Why do I feel like I need to reverse this

[stechert @stechert]

@RealJakeBroe One related meme is “your pull out game is weak”. maybe you were originally saying something to the effect of “Russia, you aren’t making progress. It’s time to pull out but your pull out game is weak.”?

[stechert @stechert]

Is it just me or does gemma 4 not know it's gemma 4? Gemma 3 knew it was gemma 3...

[Ars Technica @arstechnica]

Google bumps up Q Day estimate to 2029, far sooner than previously thought https://t.co/znyY8arBAz

[Anish Moonka @anishmoonka]

Every time you get a cancer biopsy, the lab makes a tissue slide that costs about $5. It shows the shape of your cells under a microscope, and every cancer patient already has one on file.

There’s a much fancier version of that test called multiplex immunofluorescence (basically a protein-level map showing which immune cells are near your tumor and what they’re doing). It costs thousands of dollars per sample, takes specialized equipment most hospitals don’t have, and barely scales. But it’s the kind of data oncologists need to figure out whether immunotherapy will actually work for you. Right now, only about 20 to 40% of cancer patients respond to immunotherapy, and one of the biggest reasons is that doctors can’t easily tell whether a tumor is “hot” (immune cells actively fighting it) or “cold” (immune system ignoring it).

Microsoft, Providence Health, and the University of Washington trained an AI to analyze the $5 slide and predict what the expensive test would show across 21 different protein markers. They called it GigaTIME, trained it on 40 million cells in which both the cheap slide and the expensive test coexisted, and then turned it loose on 14,256 real cancer patients across 51 hospitals in 7 US states.

The results landed in Cell, one of the most selective journals in biology. The model generated about 300,000 virtual protein maps covering 24 cancer types and 306 subtypes. It found 1,234 real, verified connections between immune cell behavior, genetic mutations, tumor staging, and patient survival that were previously invisible at this scale. When they tested it against a completely separate database of 10,200 cancer patients, the results matched up almost perfectly (0.88 out of 1.0 agreement).

Nature Methods named spatial proteomics (mapping where specific proteins sit inside your tissue) its Method of the Year in 2024, and specifically cited GigaTIME in a March 2026 update as a model that “democratizes” this kind of analysis. The full model is open-source on Hugging Face. Any cancer research lab with archived biopsy slides, and most of them have thousands, can now run virtual immune profiling without buying a single piece of new equipment.

[stechert @stechert]

@simonhg :-(. What happened now?

[Anthropic @AnthropicAI]

New Anthropic Research: Disempowerment patterns in real-world AI assistant interactions.

As AI becomes embedded in daily life, one risk is it can distort rather than inform—shaping beliefs, values, or actions in ways users may later regret.

Read more: https://t.co/gyMB2AtOuq

[stechert @stechert]

@NYCSanitation Please give us a place to report chronic offenders…some businesses absolutely refuse to do their part in any way.

[NYC Sanitation @NYCSanitation]

NYC, now it's your moment to shine. Or, shovel!

The snow has officially stopped falling. Property owners MUST shovel their sidewalks. We will issue violations for those who fail to take action starting at 12:30 pm.

[Schneier Blog @schneierblog]

Chinese Surveillance and AI https://t.co/6zxeEjReIG

[stechert @stechert]

@adcock_brett +1. The most important thing I learned from doing robotics in the real world years ago was that the real world does not have breakpoints, step, or inspect.

[stechert @stechert]

@esrtweet While any attempt at this would have flaws we can foresee, has anyone attempted to compare the rate of useful work done with and without CoC’s? How often do otherwise helpful contributors reduce contribution because of a CoC? Or because of lack of a CoC?

[stechert @stechert]

@simonhg What if we can’t tell

[François Chollet @fchollet]

The 3rd edition of my book Deep Learning with Python is being printed right now, and will be in bookstores within 2 weeks. You can order it now from Amazon or from Manning.

This time, we're also releasing the whole thing as a 100% free website.

I don't care if it reduces book sales, I think it's the best deep learning intro around, and more people should be able to read it.

[stechert @stechert]

@Austen My wife and I joined in with the echolalia when our kids were small and now we still do it at each other 🤣. “Oh ñoooo!”

[stechert @stechert]

@yacineMTB food industry:ultraprocessed addictive snacks :: social media:ultraprocessed addictive shorts? insulin resistance :: ?

[stechert @stechert]

Is it now easier to design a CAPTCHA that only AI's can succeed at than one that only humans can succeed at?

[Jack Morris @jxmnop]

the trendlines indicate that one day this app will be overrun by hordes of bots producing low quality difficult-to-identify slop

and at this point someone will be forced to build the first true Humans-Only social network. No Bots Allowed. Fingerprint Required

[stechert @stechert]

@JRIngallinera Do you know what an ASP baton is? They were illegal in California until last year, when the ban was ruled unconstitutional. In the spirit of proportional response, I’d humbly suggest that, with proper training, this could be a good, hopefully nonlethal deterrent to this guy.

[stechert @stechert]

@Suhail I wonder what fraction of work gets done before and after the regular workday at most companies. I always found those times to be the most productive personally and also when the people present tended to be the most respectful of each other's time.