Maria cecilia Jordan Anna Mills Matthew Kirschenbaum Dr. Sarah Elaine Eaton🇨🇦 I think we should focus on making our assessments meaningful and relevant to students so they feel intrinsically motivated to do the actual work, research, reading, writing. And help students find their voices, value their voice. And support them better, give them time to do it

Maria Jelmini om effektiv inlärning. Att tvingas plocka fram information via små prov, diskussioner, egna frågor osv är väldigt effektivt. Mycket effektivare än att läsa en gång till. 1/5 #rEdScand23 #rEdhan

B/c they keep popping up in my feed: I think maybe the people who helped create ML as it exists today, uncritically developed a vocabulary suggesting sentience, at best, argued that experts on bias were being hypersensitive, should maybe sit the discussion out.

People be like "AI will eliminate jobs" when what they really mean is "AI will facilitate the relentless centralization and accumulation of capital by further stealing past alienated labor and repackaging it as automation."

There's a newish genre of LLM paper I'm starting to see, where the authors put together an enormous suite of benchmarks, test a bunch of models on them, and write a 50-100 page opus talking about how LLMs can now do shiny new thing. >>

Google’s vision for the future of search shows how generative AI is all about increasing corporate power. Instead of sending you to different websites, Google has scraped the open web to generate plagiarized answers to keep you looking at ads on Google. disconnect.blog/p/google-wants…

Google’s vision for the future of search shows us how these companies are really wielding their power as they distract the public with AI hype. We need to ignore the fantasies of intelligent machines, and look at the real threats before it’s too late. disconnect.blog/p/google-wants…

1/ #Critical AI covering Meredith Whittaker’s podcast with Slate, “Who is afraid of AI?” As the tagline says:  “Real harm from A.I. doesn’t have to look like Skynet—it’s already happening.” Let’s listen! slate.com/podcasts/what-…

/8 Chatbots simulate familiar responses. It takes resistance to not “anthropomorphize” those simulations: these models are not sentient. Without understanding ppl don’t recognize the limitations of so-called artificial intelligence. “AI is not a Frankenstein.”

/11 Only the biggest companies can produce these systems. Tech isn’t being produced or distributed equally or democratically: it will perpetuate ongoing marginalization.

16/ “AI is not magic, it is based on concentrated resources and concentrated computational power, concentrated data resources, generated by surveillance and the concentrated power of these companies. “

Lukk øynene og tenk på øyeblikket der Sverigedemokraterna innser at de nå må bla opp enormt mye penger til et digert internasjonalt kulturarrangement med tydelig Pride-preg i 2024

Tech companies are out here claiming that high-tech generative artificial intelligence is the reason they are laying off workers yet behind the scenes are actually hiring low-paid human labor but then generative AI will get the credit.

Tror det må sies en gang eller to om dagen: ChatGPT er ikke intelligent per menneskelige trekk, den er bare sjukt godt på å gjette neste ord i en rekke. Inga Strümke

Christopher Newfield John Warner Anna Mills 1. Educate yourself/your colleagues about tech/political economy. 2. Teach your students about "AI." 3. Organize your syllabus & assignments w/ the understanding that many of your students will be using chatbots - clarify your own policies. 4. Use AI in classroom in limited way.

Christopher Newfield John Warner Anna Mills 5. Don't force students to sign up for subscriptions. 6. When using "AI" consider using it for the kinds of probing experiments that eripsa (professional robot) has developed from Deb Raji's templates & which enable students to act as researchers, probing errors, stereotypes etc. /9

That's verging on misinfo. "Neural nets" are called that because they are built out of software components roughly inspired by a 1950s idea of how neurons work. Promoting the narrative that they are "like brains" insidiously promotes the idea that ChatGPT et al are "thinking".>>

@ShannonVallor Ppl are constantly misled into believing that LLMs are "training" "learning" in real time (as humans do) in ways that produce radical misunderstand. But these are pre-trained models (PT in GPT) & retraining is hugely expensive. "Fine-tuning" also not a small deal. /2

Arvind Narayanan /b Chatbots and language models were not designed for educational purposes & no educators were consulted. Yet industry & some journos now expect educators to leap to implement as if these are godsends for teaching, learning, and assessment.

Arvind Narayanan /c Sez who? Where are the peer-reviewed studies that students learn anything through these particular tools. Perhaps they'll emerge but in the meantime, let's get off the back foot and not be cowed into making chatbots the center of our classroom strategies.