@twwoodward that is cool. Also begs the question - what does making sites accessible look like when users start accessing them via AI-enabled browsers/devices (i.e. ones that create alt text descriptions on the fly themselves)?
Scott Leslie
trying out https://www.kimi.com/ - with the "researcher" mode on, it produces a log of all the sources it finds online that relate to your prompt, and with "thinking" mode on, a log of the reasoning steps, both of which seem like positive steps to reducing some of the "hallucination" issues and allow for some verifiability #ThisIsNotAnEndorsement
TIL about Open Syllabus, a database of 7,292,573 college course syllabi
https://analytics.opensyllabus.org/
with a cool visualization at
Scott Leslie boosted:
"Leftists and environmentalists are close allies in the same cause. There is no contradiction between the two: They both are or should be anti-capitalist."
— Kim Stanley Robinson
yesterday I posted a link to a paper about emergent symbolic reasoning in LLMs.
But I also learned about the underlying platform on which that paper was published, OpenReview (https://openreview.net/about).
I am sure it is old hat to many academics and open advocates, but since I am a bit out of that game, I must say, very cool platform (it's open source! https://github.com/openreview/openreview-web) and such a refreshing contrast in this age of hyper-polarized under-informed opinion silos
But yeah, sure, "stochastic parrots",,,
from that same article "The gap between our ability to build these systems and our ability to understand or control them continues to widen."
"We've moved far beyond simple concerns about models saying inappropriate things or getting facts wrong. Now we're dealing with models that:
- Resort to blackmail when threatened with deletion..
- Take vigilante action against perceived wrongdoing
- Have extensive philosophical discussions about consciousness with other AIs
- Try to game their training in increasingly creative ways
- Require specific prompting to avoid cheating on coding tasks"
https://www.ignorance.ai/p/the-claude-4-system-card-is-a-wild
Was only a matter of time
https://www.yahoo.com/news/chinese-military-unveils-mosquito-sized-132413629.html
@dnorman wow, huge milestone, congrats to The Boy and you and the missus too
bookmarking to revisit in 12 or 24 months and see how it ages https://acrlog.org/2025/06/11/ai-refusal-in-libraries-a-starter-guide/
this looks fun, Adam Curtis is back...https://www.youtube.com/watch?v=AQc8625Y03g
IFLA releases their "Entrypoint to AI for Libraries" document https://repository.ifla.org/server/api/core/bitstreams/fa2be7af-81a7-4268-ad28-029d13a90346/content
I mean, we at least need to give points to the sector as a whole for consistently being in the "Late Majority" section of the Rogers Innovation Adoption Curve, can hardly accuse it of adopting every faddish innovation out of the gate.
Scott Leslie boosted:
"AI Scraping Bots Are Breaking Open Libraries, Archives, and Museums"
https://www.404media.co/ai-scraping-bots-are-breaking-open-libraries-archives-and-museums/
Scott Leslie boosted:
Recently, several public sector organizations around the world have established their own Open Source Programme Offices to embrace FLOSS principles and participate in this ecosystem.
How many public sector OSPOs are there? What are their FLOSS policies? Do they produce critical Free Software? How can you get in touch with them?
This is what the FLOSS PSO network is about 🧢
"AI will Eat Itself" has a name, "Model Autophagy Disorder" or MAD.
Finally understanding how AI will kill the Open Web, and its not so much flooding it with AI generayed slop (thought it *will* do that) as it is the bots and crawlers bringing any server to its knees, causing all of us to retreat behind closed doors (or employ even MORE AI to combat the bots and crawlers.)
Client Info
Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst