Does AI help or hurt programmers? Conflicting studies can be explained by "technical debt"—quick fixes that lead to long-term headaches. Nor will shifting their job from writing to reviewing AI-generated code make devs any happier www.linkedin.com/posts/jonipp...
The National Institute of Standards and Technology has had enough of your IT department's BS password requirements. Long random phrases are more secure than ridiculous combinations of numbers and letters that you have to change every three months arstechnica.com/security/202... 🐘♻️
Proposed guidelines aim to inject badly needed common sense into password hygiene.
Does generative AI build models of the world or just average what's in its training data? Image generators add weird average JPEG-like errors into images that shouldn't have them, because they can’t tell intentional details from accidental artifacts www.linkedin.com/posts/jonipp... 🐘♻️
Does generative AI build models of the world or just average what's in its training set? It appears that even the errors in AI images are averages. Most of us…
Can traditional assessment survive AI in the classroom? I argued no in today’s Times Higher Education panel organized by @mahabali.bsky.socialwww.timeshighered-events.com/digital-univ...
Join us in Cairo, Egypt, for Digital Universities Arab World 2024 to explore how new and emerging technologies will support a future vision of digital-first higher education.
A shock to see my sleepy hometown of Springfield abruptly become the target of a racist disinformation campaign. I've been monitoring AI election threats for the past year but did not have "AI-Generated People in Ohio Eating Pets" on my radar www.nbcnews.com/tech/misinfo... 🐘♻️
The ChatGPT group "remembered nothing, and they all failed"—MIT computer science prof Eric Klopfer reminding us that learning requires friction. (I just wish Klopfer also assessed improvements in prompting technique.) cacm.acm.org/news/the-imp... 🐘♻️
Read this Stanford paper making the rounds closely and you'll see that generative models do better at combining established ideas than true boundary-pushing creativity, thanks to the way averaging can surface "in-between" research opportunities. More at www.linkedin.com/posts/jonipp...
This Stanford paper concluding AI produces novel ideas better than humans has attracted both acclaim and criticism (see Eryk Salvaggio for the latter). But I…
The advent of generative AI could transform the role of the archive into something dramatically different than it was just two years ago. Here's a 7-minute video I made for my digital curation students on how this could play out youtu.be/rQ54XZREStI 🐘♻️
YouTube video by UMaineDigCuration
Why do AI outputs often feel "average"? Ted Chiang argues LLMs simply average other writers' choices, which sounds like oversimplifying—but averaging turns out to explain AI's brilliance and occasional nonsense. I explain at blog.still-water.net/why-your-ai-... 🐘♻️