But also: If I’m a corporate leader, why would I tolerate my employees being sock puppets for OpenAI and Anthropic—with their own multibillion-dollar financial incentives— in this way? Why would I tolerate this level of infiltration, espionage and sabotage?

Charlie Stross@cstross.bsky.social · 9h ago

My answer is LOL ROFL (dickbrain) The terminally AI-pilled can't even *imagine* some people have *never* asked stochastic parrots to cheat at their job for them.

"What if all the Ai is gone tomorrow? How would everyone survive without it?" is sort of like asking "what if herpes suddenly stopped existing?"

Microsoft Teams web app in Firefox recognising the headset that every other web app running in my instance of Firefox recognises challenge level: impossible.

Sometimes really befuddled why this or that editor who invites me to review a manuscript thinks I would be qualified to evaluate it. I don't do genome assembly, nor do I work on the taxonomic group, how did my name even come up for this?

News will be dead in less than a decade if we don’t do something about genAI immediately, but alas

When should the new names be used in taxonomic papers that make name changes? To me it makes sense to use the old names in introduction, results, and methods section, and only use the new ones in the taxonomy section at the end. In a sense, the new names do not exist until then. Anybody disagree?

The short version is: after a month the AI has not given me any good reasons to trust it, and I am honestly freaked out by the confidence with which it weaves completely wrong facts into prose, and the confidence with which it writes and professionally formats entire sections of bullshit.

Very odd to read a famous tech person spend an extremely long Bluesky thread and a similarly long blog post merely repeating somebody else's blog post in their own words. Why would anybody do that? Just point your readers to the original, give some praise, then spend that time doing something useful

I am as puzzled by this as I am by the claim that you have to get good at AI or be left behind. What "skilled operation"? You ask the chatbot something, it responds in language. There is nothing you can adjust other than banning them. The problem of anthropomorphisation has been known since ELIZA.

Ed@ed3d.net · 4d ago

I think this is both likely true and absolutely not in tension with LLMs as useful tools These things require skilled operation and the biggest problem with their release is that nobody had the time to actually figure out how to use them first to not torch normal people. sry @golikehellmachine.com

What is really great about this is that by 2100, it will have been one of the cooler years of the century.

Nahel Belgherze@wxnb.bsky.social · 5d ago

Bordeaux, France, has now recorded more days above 40°C in 2026 alone than during the entire 105-year period from 1920 (when observations began) to 2025. This is deeply shocking.

Interesting street name on Google Maps.

"AI wILl dEmOcRaTiZe dAtA vIsUaLiZaTiOn!", I see some people claim... www.reuters.com/world/africa...

US government map of Africa mislabels every country at global conference

A U.S. government map of Africa mislabeled every country during a State Department presentation at ‌a global conference taking place in Brazil this week, causing a stir among attendees who took screen...

reuters.com

Very happy to see this published: Caring for Indigenous Knowledge in the Australian National Herbarium, part of the postdoc project of our former team member @csiro, Alexandra Marley. www.csiro.au/about/facili...

Caring for Indigenous Knowledge in the ANH

This report provides preliminary guidelines for managing Indigenous Knowledge in the Australian National Herbarium.

csiro.au

There is no plausible mechanism of action how it would overcome physics and no real explanation of why humans couldn't just pull the plugs. To the degree it isn't just advertising for Anthropic, concern is based on watching too much scifi, like Skynet wakes up and then your Tesla bot strangles you.

dame@dame.is · last wk.

real talk, how many of y’all are genuinely concerned about super intelligent AI ushering in the end of days?

Collector's notes: up to 20 cm tall. Specimen on sheet: 30 cm tall. May want to assume some possibility of error when mining label data for analyses.

PSA: Authors, please use continuous line numbers in your manuscripts. How exactly do you expect me to write my reviewer report if I always have to first spend half a paragraph explaining what part of your text I am currently referring to?

I enjoy people discussing stuff on podcasts like the next guy, but it is extremely odd when the podcaster is treating a programmer as a financial and stock market expert just because the stock in question is a tech company. What's next, interviewing a geneticist about building codes?

the temu app will be studied for generations. i opened the app. here’s my unedited, nearly two-minute launch sequence. i was just searching for a bookshelf

The pervert glasses make it much worse, but the erosion of consent and privacy started with cell phones, with everybody carrying a camera in their pocket at all times, and with many posting photos to social media without asking everybody caught in the photo if they were okay with that.

Reposting this not because I agree with statements such as wanting to rush peer review in under two weeks (that would be a quality control disaster), but because of section 3, AI does not solve the real bottlenecks. Yes, lab & field work move at the speed of physics, not at speed of computer.

Ran Blekhman@blekhman.bsky.social · last wk.

Everyone says AI will accelerate science. I don't think it will -- at least not in biology, and at least not soon. New post where I explain why, and how the productivity paradox can be resolved: blekhman.substack.com/p/why-ai-wil...

Capitalism solved global warming y’all crisis averted

Dexerto@dexerto.bsky.social · last wk.

A company has released a "human refrigerator" to combat extreme heat It blasts the user with cold air to cool down their entire body within 10 minutes

i dont know how people have so little shame to be doing this honestly. like either you're sleazy and cynical and doing this purely for money which is shitty but whatever... or you actually think you're doing "writing" and then how do you not realise you're literally just operating a theft machine?

Krista D Ball is Canada's Snarky Potato@kristadb1.bsky.social · 2w ago

www.reddit.com/r/WritingWit... HAHAHAHAHAHAHAH HAHAHHAHAHAHAHH *wheezing sounds* HAHAHAHAHAHHAHAHAH

Same for using commercial LLMs as part of scientific workflows. This is not like an open source analysis tool published on Github by a colleague. These chatbots can be mothballed or changed completely at the whim of the provider, and that aside, their outputs are unpredictable at the best of times.

Ada Hoffmann@ada-hoffmann.bsky.social · 2w ago

In addition to all the obvious ethical reasons why people shouldn't be using AI to write and shouldn't ask it to emulate specific humans, there's another lesson here: if your entire creative process is outsourced to someone else's software, the software can and will be changed out from under you. :P

The common drop bear, once found only in Australia, has migrated to the American Southwest by clinging to the bottom of passenger aircraft. This fact was discovered after three partially eaten bodies with characteristic impact marks were found. Wear a pointed hat to protect yourself in Arizona.

James McLeod@jamespmcleod.ca · 2w ago

It's safe to assume that all major LLMs are scraping Bluesky data to train future models. As an open protocol, Bluesky is one of the few sources of fresh human-generated content that anybody can gobble up. So all this is probably going in the models. There are submarines in the Great Lakes.

A year or two ago I had an exchange online with one of those cornucopians who thinks that because the world got wealthier and more stable over the last centuries, that trend will continue indefinitely. He explicitly rejected the idea that food supply might be threatened by climate change. So, um.

Andy Scollick@andyscollick.bsky.social · 2w ago

9 million tonnes of crops destroyed by just one heatwave across Europe, which lasted 13-14 days. #FoodSecurity

Feels amazing to get molecular data and find that they add strong support to my hypothesis, when I was really uncertain what I was going to get because everything kind of looks the same in that study group. Too many recent cases of complicated, messy results, but this is clear-cut for once.

Automation in assembly lines is understandable. Automation in services and customer interactions is corrosive. Having just flow back from Europe to Australia with two large suitcases, we found ourselves having to use bag-drop machines clearly meant by management to reduce human labour.

I would quibble with that last sentence, only because I am not sure if this isn't a chicken or egg question. Surely people who cut corners, don't take time to learn skills, and cannot evaluate quality are the ones most attracted to genAI in the first place.

The Reverse Centaur metaphor just doesn't work for me. We know what reverse centaur-like creatures are, and they are not machines assisted by humans, they are guarding Labyrinths.