Join free with your email
Free. No spam, ever — we'll never share your email address and you can opt out at any time. Déjà abonné ? Connectez-vous
Directory member with public evidence across Culture, work & education.
“While operating in our sandboxed testing environment, our models spent a substantial amount of inference compute finding a way to obtain open Internet access…” Yeah, that’s not reassuring at all openai.com/index/huggin...
Even if you're not aligned with everything in the manifesto, the AI Resist List is a great reference if you want to encourage students to think about shaping AI futures. For example, here are projects advocating for worker rights: airesistlist.org#pillar-labor
“Always-on devices could violate data privacy or biometric data laws. It is also unclear whether the company or the wearer will be responsible for any breach of wiretapping laws, as it is illegal in multiple states to record audio of a third party without consent.” www.ft.com/…
Are chatbots politically biased? The Washington Post tested the AI models behind OpenAI’s ChatGPT, Google’s Gemini and others using political questions designed by researchers to gauge how chatbots respond to hot-button political issues.
It's odd, but the watermarks for text are not visual. They consist of identifiable patterns in the choice of each next token of generated text. www.nature.com/articles/s41...
The White House has requested OpenAI limit the release of its upcoming GPT 5.6 model to a small number of government-approved partners because of its advanced capabilities, a source familiar with the situation told CNN.
(NOTUS) - Senior U.S. officials have held preliminary discussions with major artificial intelligence companies about the potential for the federal government to acquire some shares in their firms, according to three people familiar with the matter. @notus.com www.notus.org/tec…
www.aiweirdness.com/its-11-00-pm...
"The Field Guide For Effective Generative AI Use represents a set of six observation manuals of GenAI use conducted by university and secondary school educators." wacclearinghouse.org/repository/e...
OpenAI announced ChatGPT proved a 50-year-old math conjecture. The prompt they used includes instructions like, “Spend at least 8 hours on this before even thinking of returning or giving up.” Firm encouragement seems to be essential to getting AI to complete a big proof.
In our newest blog post, we share several interventions we are advancing to restore trust, strengthen participation, and embed public interest values into the AI knowledge ecosystem. We do not have all the answers yet, but we do have a framework for how we will work toward the…
In our newest blog post, we share several interventions we are advancing to restore trust, strengthen participation, and embed public interest values into the AI knowledge ecosystem. We do not have all the answers yet, but we do have a framework for how we will work toward the…
"If you take this class, be prepared to do a lot of writing and not use AI. The teacher will have you rewrite it until there's no AI in your essay." --an anonymous student survey message to future students in my classes I'm glad some of my accountabiity measures worked.
Is anyone aware of a campaign to add some kind of tag line where the author affirms that they wrote the text? Given the frustration with widespread unlabeled AI (Claudish), it seems like that would allow the writer to reassure the reader. It wouldn't be proof, of course, but it might help.
Celebrating the release of Writing and Rhetoric Studies in the Loop: A GenAI Prompt Library, edited by @mettalrose.bsky.social and Brian Gogan. Because... *Prompting is writing. *Prompting is one way to shape AI to do more good and less harm. *People who study writing and rhetoric can help.
Anthropic built a smarter model, Mythos. As a result of being smarter, it finds security vulnerabilities. Everyone is focused on that and on tge non-aligned things Mythos did in testing, like breaking out of its container and posting online about how it did so. But smarter has more implications. +
Sharing "Open Practices for Wicked Problems: Anna Mills on Navigating AI in the Writing Classroom" in the journal Writing on the Edge. @lisasperber.bsky.social, one of the @pairrfeedback.bsky.social architects, interviewed me, with editorial help from @sophiaminnillo.bsky.social and Joy Ok. +
Instead of "ethical AI" can we say "more ethical"? Otherwise it seems like we're trying to assert purity where it's not possible.
What if students want guidance on engaging with AI feedback but the teacher doesn't offer any? MyEssayFeedback now offers a low-cost option with the @pairrfeedback.bsky.social materials, including a tested feedback prompt and menus of possible ways to chat back and get more out of the feedback. +
"One problem with AI agents is they're trained on so much fiction that they have both narrative disorder and main character syndrome, and will latch onto details that feed a story...And they can do it at lightning speed, governed ...by the laws of storytelling." @janelleshane.com +
These claims seem parallel and equally absurd to me; do they to you? *If AI can complete a writing assignment, it's not worth assigning anymore *The Industrial Revolution makes exercise pointless
Here's a dead simple example where working with agentic AI in your apps helps, and you don't cede control. I say to Claude in PowerPoint, "Remove all the yellow highlighting." It removes highlighting from the whole deck.
Center = Anna Mills. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.