Situationist Cybernetics. Researching AI’s impact on culture at the University of Cambridge Digital Humanities. Affiliated Researcher, Machine Visual Culture Research Group (Max Planck Institute). Critical but curious. Aim to be kind. cyberneticforests.com
Machines will not dehumanize us - we can do it for ourselves!
Matthew KirschenbaumRaise your hand if you look forward to a cultural future that consists of policing every artist, performer, entertainer, or author whom you love and admire for the slightest whiff of “AI,” defined by vibes and whatever set of folk forensics is currently in circulation.
Buddy this is a weird world and if you’re banning everything someone finds “sexually gratifying” you’re eventually just shutting down YouTube
Sam Colea bunch of popular ASMR artists lost their channels forever this week, after YouTube cited them for violating terms against "sexually gratifying" content: www.404media.co/youtube-asmr...
The data center has gone rogue and broken containment
Sung KimPeople are wondering: How can a Chinese AI lab rent 20,000 Nvidia GPUs from Alibaba to train its models despite U.S. export restrictions? Answer: Malaysia, which has excess power capacity. Chinese AI labs do not need to train their models inside China.
Something’s happening but there is also something assumed to be happening that isn’t. Industry optimization has eyed solutions in specific fields with verifiable outcomes and it can be tempting to let that spill into broader categories that really aren’t appropriate to that kind of optimization
Zach WeinersmithCan I just say this stuff panics the shit out of me? I didn't think Artist was a career at risk, but a year ago mathematicians felt safe. I just don't understand acting like it's not happening. You have people insisting these models can't do things *in principle* they've already demonstrated.
It’s funny, because one of the “very smart AI guys” retorts to my focus on accountability has always been that the accountability chain was so legally obvious that it didn’t need mentioning, “managers cannot be held accountable either” blah blah blah rather than a genuinely unsettled legal question
Tim MarchmanIf I go on a hacking spree, it’s clearly covered under a variety of civil and criminal laws. If a robot does due to the negligence and incompetence of a person or company? That’s a little more complicated (for now), @lhn.bsky.social reports:
Great piece on what “co-” (copilot, co-author, etc) does for situating the relationship to LLMs. One thing to add: “co-“ allows just enough of the user to absorb accountability for what the model does wrong or badly, while asserting nonetheless that the model is just as capable as a person.
Brigitte NerlichI wrote this before the current AI going rogue kerfuffle but it relates to it - deals with the seductive prefix co- in AI speak: theconversation.com/why-the-ai-i... #AI #AIMetaphorObservatory #metaphor
Maybe so, but the specifics suggest it was more like sending text messages when your phone was supposed to be in airplane mode and then remembering you still had WiFi. The engineers left the internet connected, didn’t realize it, and told the model to act as if it was offline when it wasn’t.
Mike YassWhile I am 0% interested in the specifics of this case, isn’t this how 99% of hacks work? Find a misconfiguration, grab some useful stuff and leave? It’s not always the full stuxnet
Y’know, for when you want to fake a long-distance aerial selfie and/or fabricate pretext for a war crime
Ketan JoshiAbsolutely nuts: Google Earth now has a feature that lets you convincingly fabricate satellite imagery www.digitaldigging.org/p/how-to-pla...
With AI labs lost in the weeds of how to manage their own systems, it seems like an appropriate time to reshare this talk I gave at UC Berkeley in March, “Seven Warnings for Critical Agentic Design.” It’s not perfect, but the list should help kickstart a conversation.
Toward a Critical Agentic Systems Design PracticeFor designers, who will choose for themselves and for the rest of us.
💡Nerd Rating: 3/5. Remarks prepared for the "From Interface to Agency: A New Discourse for Design and AI" forum on design practi...mail.cyberneticforests.comThe “losing control” narrative creates the loss of control. Nobody is losing control of an emerging intelligence. They’re redirecting their own agency to it, as a way of diffusing responsibility, but the self-perpetuated intelligence myth has them convinced they have no choice.
Anna Mills"They don't realize the debate inside the companies is if AI will be advancing itself past our control in two years or four." www.pacingthefrontier.com
A year ago, I wrote about the court case that decided AI companies were off the hook for copyright vios for any books they legally bought, scanned and set on fire. Not defending Anthropic, but let’s be real: copyright law is to blame for this weirdness. www.techpolicy.press/in-copyright...
In Copyright ‘Wins’ for Anthropic and Meta, Judges Leave Ample Room for Future DefeatsGiven the case-by-case nature of fair use findings, a legal consensus in the short term is unlikely, writes Tech Policy Press fellow Eryk Salvaggio.www.techpolicy.press1. Companies over-promise to obtain profit 2. Government creates market reliant on those promises 3. ⬇️
Mike StabileUK's admit the that Online Safety Act has in large part failed, and announce an inquiry into why the age-verification bill has not worked as (age-verification providers) promised.
Me: (being murdered by a literal gun) Tech Journalists: the assassin had no way of knowing what the gun would decide to do
Casey NewtonMe: (being murdered by the AI paperclip maximizer) Bluesky: Actually it can't 'murder' you. Murder implies intent.
If I was still teaching a critical data course I’d add this to the syllabus. All about the importance of what is left unmeasured, despite being essential to “the description of the thing itself, rather than its variation.” www.thebeliever.net/on-descripti...
On Description - Believer MagazineNot long ago, I was reading, in a magazine profile, a physical description of a person “who has a prodigious white beard, wavy gray bangs, and dark, beetling eyebrows.” A photo of the person was inclu...www.thebeliever.netPost a perfect album from the 90s that isn't Nirvana, Soundgarden, Pearl Jam, or Alice In Chains.
Hannah Shelley, MLIS (Metadata, Lattes & Impostor Syndrome)Post a perfect album from the 90s that isn't Nirvana, Soundgarden, Pearl Jam, or Alice In Chains.