Scientists found something surprising about #AI chatbots like #ChatGPT.

The Dunning-Kruger Effect doesn’t work the same way with AI.

When people used ChatGPT, nobody was good at knowing how well they did.

Everyone thought they did better than they actually did.

The scientists also found something backwards from the usual effect.

People who thought they knew a lot about AI were the most wrong about their abilities.

They thought they were better at using AI than they really were.

2/2

Bevel, a New York-startup that has become one of the fastest-growing health apps in the U.S., has raised $10 million to scale its AI health companion. Here’s more from @Techcrunch on what the company does and where it’s headed:

https://flip.it/6AVvpa

#Tech #AI #Startups #Technology #ArtificialIntelligence

Wired: No, SNAP Benefits Aren’t Mostly Used by Immigrants
SNAP benefits are set to run out on Saturday. Far-right influencers and extremists are incorrectly claiming that immigrants are the main recipients of food stamps.

"...A number of conspiracy theorists and right-wing influencers are claiming that immigrants are the main recipients of food stamps, while AI-generated videos on TikTok push racist stereotypes of Black people demanding more benefits...."

https://www.wired.com/story/disinformation-conspiracy-theories-snap-benefits/

#AI #disinformation #snap #hunger #racism

Alt...

Wired: No, SNAP Benefits Aren’t Mostly Used by Immigrants SNAP benefits are set to run out on Saturday. Far-right influencers and extremists are incorrectly claiming that immigrants are the main recipients of food stamps.

Language Models are Injective and Hence Invertible https://lobste.rs/s/ukvfgs #ai #math #security
https://arxiv.org/abs/2510.15511

More or fewer devs because of AI?
Jim Zemlin (@linuxfoundation) told The New Stack:
AI isn’t replacing developers—it’s expanding their impact.
🎥 Full interview: https://www.youtube.com/watch?v=X1c3hGfF_a0&t=12s
#AI #OpenSource #Developers

Clarivate Releases Pulse of the Library 2025 Report https://www.infodocket.com/2025/10/30/clarivate-releases-pulse-of-the-library-2025-report-2/ #libraries #AI #GPT

Report: Texas #Schools are Using #AI to Screen Library #Books Under New State Law https://www.infodocket.com/2025/10/29/report-texas-schools-are-using-ai-to-screen-library-books-under-new-state-law/ #libraries

Denmark seems to be pursuing the idea of protecting people from "AI" deep fakes by addressing people's image as part of copyright law. [1]

I applaud the idea of doing something about this, and this is a better approach than none at all, but it's not quite how I would pursue it.

For one thing, there are a number of places where people are innocently captured in images that this will create complications for. And for another, I don't think it's powerful enough to address the real problem.

The "MOO" community (MOO is MUD, Object-Oriented, and MUD is Multiple-User Dungeon, and Dungeon was one of the first text-based interactive fiction games, also called Zork), this came up a long time ago. Ironically, since MOO is entirely text, images were not involved. But there was still the issue of appropriating people's view of themselves for ill purposes, and this was richly discussed.

MOO, which had its greatest popularity in the 1990's, before Second Life overshadowed it, functioned as a kind of textual sketch of things to come. It was a coarse level of detail because its technical layer doesn't allow for super-elaborate detailing, but that forced the social aspect to be the focus rather than the technology. Modern systems purport to capture reality, but they often get so side-tracked on making things photo-real visceral experiences that they give short shrift to the full complexity of human social interaction. So they're still catching up to some of the social issues MOO explored decades ago.

In Julian Dibbell's fascinating book "My Tiny Life: Crime and Passion in a Virtual World", which you can and should buy if you can afford to, but which the author arranged to be freely downloadable as a PDF for those who could not afford it [2], the focus is on "A Rape In Cyberspace", he explores some of these issues. Originally published in The Village Voice and later adapted for a book, this story is, in the author's words, "a True Account of the Case of the Infamous Mr. Bungle, and of the Author's Journey, in Consequence Thereof, to the Heart of a Half-Real World Called LambdaMOO".

This story will not tell you how to understand what Denmark is doing, but I think it informs my way of thinking about this issue.

At its core, both situations--the issue in cyberspace and the modern issues in the real world--are not "infringements" (in the way copyright would talk about them) but "violations" in the way a person's sense of self matters.

Some people will point to rape as a matter of physical violation, but just as others are quick to say it's not a crime of sex, it's a crime of violence, I would similarly say it's a crime of violation, of taking control of a person's sense of self. And that's what's in common with these other matters, like grabbing someone's image.

We don't presently have a standard for this, and like many matters of human endeavor, there is extraordinary nuance. Fair use, one might say. Certainly parody is one place where people don't have complete say. The sitting President wants to go after critics for disparaging his good name, for example, and ordinarily the disparaging of someone's good name might be seen as a violation, but in certain realms of public discourse, especially for public figures, we allow and insist on it.

This is partly true, too, because even underlying the issue of violation is the issue of power. The law is really at its core protecting those powerless to protect themselves. So, for example, while it might be a violation to appropriate the good work of an actor who's just struggling to eat, selling their image royalty-free, appropriating the name of a politician who can with the stroke of a pen cut the food supply of millions is not exactly exerting power over them, certainly not unconditionally dominating power.

So we should be careful in our understanding of good law to understand that it seeks not a bright line of pain to itself become a weapon, but rather just an ability to tip power balances back toward the middle, making the world an even battle among people who are born into different levels of power and who cannot therefore fairly be expected to solve their own problems.

I've swept through many issues here, but I have decades of thought underlying my reaction to Denmark's idea, informed by the lucky accident that I was there at the time LambdaMOO sketched the future.

I sometimes note in conversations with people for whom a topic is new and hypothetical that they will say "I wonder what would happen if..." and I reply in the past tense saying "Oh, this is what happened." Because I don't have to speculate. I saw it. It mightn't happen reliably that way again. Many possibilities were in play. But even those were tangibly close to my experience. I have rich, detailed thought because I lived at least one version of it. Just wanted to share that.

[1] https://www.weforum.org/stories/2025/07/deepfake-legislation-denmark-digital-id/

[2] https://epdf.pub/my-tiny-life-crime-and-passion-in-a-virtual-world.html

#AI #IP #Law #Crime #Copyright #DeepFake #DeepFakes #violation #rape

🎤 Upcoming at SeaGL 2025:
📍 03:00 PM on November 08
🗣️ "Local Offline AI"
👥 Speaker(s): Adam Monsen
📍 Room: Room 145
🏷️ Track: Open source AI and Data Science
📝 Join Adam's talk to learn how to stand up offline AI models for powerful and privacy-friendly advanc...

#SeaGL2025 #ai #business #privacy #data #FLOSS
🔗 https://pretalx.seagl.org/2025/talk/VD9V3K/

Continuous NVIDIA CUDA Profiling In Production via @brancz https://lobste.rs/s/keiakc #ai #linux #performance
https://www.polarsignals.com/blog/posts/2025/10/22/gpu-profiling

How non-disclosure agreements keep AI data center details hidden from Americans.

From NBC News: "Big Tech companies use secrecy agreements with local governments to keep communities from knowing who is building in their backyards."

https://flip.it/PMQyfd

#AI #ArtificialIntelligence #Technology #USPolitics

🎤 Upcoming at SeaGL 2025:
📍 02:00 PM on November 08
🗣️ "Today I Learned.... The 2025 FLOSS Research Roundup"
👥 Speaker(s): Kaylea Champion
📍 Room: Room 332
🏷️ Track: Education
📝 Friends, it's time once again to review the newest research findings about FLOSS. What's new in 2025...

#SeaGL2025 #education #research #ai #FLOSS #opensource
🔗 https://pretalx.seagl.org/2025/talk/QAZYWP/

Okay, this is dumb. Can't read the article, but this stuff wasn't around a few years ago. Can this guy NOT READ A MAP?

#AI

Alt...

NY Times article: 48 Hours Without A.I.

Alt...

Man looking confused trying to use a paper map

As #AI and automation reshape work, @amazon’s Future Ready 2030 initiative shows how tech can empower. A $2.5B investment will equip 50 million learners with future-ready skills, turning technological progress into human progress. #TechForGood https://www.aboutamazon.com/news/workplace/amazon-future-of-work-skills-jobs-training

Visual Features Across Modalities: SVG and ASCII Art Reveal Cross-Modal Understanding in LLMs https://lobste.rs/s/om8fjf #ai #science
https://transformer-circuits.pub/2025/october-update/index.html#svg-cross-modal

🎤 Upcoming at SeaGL 2025:
📍 02:00 PM on November 08
🗣️ "“Hidden in Plain Sight: Addressing Data Bias in AI-Driven Systems”"
👥 Speaker(s): Autumn Nash
📍 Room: Room 145
🏷️ Track: Open source AI and Data Science
📝 As AI increasingly powers critical systems across industries, the quality and neutrality of training...

#SeaGL2025 #ai #ml #performance #automation #data
🔗 https://pretalx.seagl.org/2025/talk/ETZQ8V/

Let's do a deep dive into this paper: "Why Language Models Hallucinate."

When this came out, many people's summary was "even OpenAI admits that hallucinations are a fundamental problem of transformers/autoregressive models/LLMs."

I've seen many people conclude that this means OpenAI is grifting, knows they're building on a wrong paradigm, knows they won't get to AGI etc.

#ai #paperdive

1/n

Alt...

The first page of the paper "Why language model hallucinate" by Kalai et al. Three of the authors are from OpenAI. Parts of the paper are highlighted.

#ChristGPT

#AI #chatgpt #grok #gemini #claude

Alt...

Technology An ex-Intel CEO's mission to build a Christian AI: ‘hasten the coming of Christ's return’ Patrick Gelsinger, executive chairman of Gloo, has made it his mission to advance Christian principles in Silicon Valley

In Acuitas news, I dust off episodic memory and try to unify it with the more recently developed narrative architectures. These open new possibilities for methods of summarizing, characterizing significance, and selectively forgetting memories.
https://writerofminds.blogspot.com/2025/10/acuitas-diary-89-october-2025.html
#AI #ArtificialIntelligence #SymbolicAI #NLP #chatbots

Alt...

A clutter of assorted books stacked on a concrete floor. "Disorganized books piling one another", by Ibrahim Husain Meraj via Wikimedia Commons. See source for license.

Ah, good thing Ireland didn’t bet a major chunk of its economy on US Big Tech creating jobs or anything.

*phew!*

#BigTech #USA #AI #tech #fascism #technoFascism #Ireland #Amazon

Alt...

Amazon has six sites in Ireland, including the Amazon Fulfilment Centre in Baldonnel Business Park in Dublin, Ireland. Amazon to cut 30,000 office jobs internationally following Al investments It is not yet clear whether Irish employees will be impacted. Today 8:40 a.m. AMAZON WILL LAY off tens of thousands of office workers as the e-commerce and tech giant trims costs amid ramped-up investments in artificial intelligence, according to US media reports. Some 30,000 positions will be cut in a belt- tightening move expected to begin on Tuesday, multiple news outlets reported.

Given that today's artificial intelligence is mostly trained on human mediocrity (and obsequiousness), I don't think it will surpass the best of human intelligence any time soon.
Or rule the world.
_________
What are you reading, my prince?
Hype, hype, hype...

#AI
#ArtificialIntelligence

Last week (somehow I missed it) a BBC-led European Broadcasting Union study reported on the reliability of AI-assistants reporting the news.

Among the key findings were:

45% of all AI answers had at least one significant issue;

31% of responses showed serious sourcing problems – missing, misleading, or incorrect attributions;

20% contained major accuracy issues, including hallucinated details & outdated information.

More enshitification!

#AI #NewsMedia #politics

https://www.bbc.co.uk/mediacentre/2025/new-ebu-research-ai-assistants-news-content

I often cite Ray Bradbury's brilliantly poignant short story "The Flying Machine" when talking about this ill-considered headlong rush for an "AI"-mediated society that is being thrust upon us.

Today I learned there's a version of that must-read short story in graphic short story format. Take the time to read and appreciate the point Bradbury was trying to make.

https://www.onverticality.com/blog/bradbury-flying-machine

H/T @stever

#ai #ethics #wisdom #society #tech #technology #bradbury #RayBradbury #TheFlyingMachine #GraphicShortStory #GraphicNovel #comics #ComicStrip

🎤 Upcoming at SeaGL 2025:
📍 11:30 AM on November 08
🗣️ "The CLI Renaissance: Why Command Lines Matter in the Age of AI and the Promise of MCP"
👥 Speaker(s): Sriram Madapusi Vasudevan
📍 Room: Room 332
🏷️ Track: Open source AI and Data Science
📝 As AI reshapes our digital interfaces, command lines aren't fading—they're evolving. This talk exami...

#SeaGL2025 #ai #cli #privacy #cli #FLOSS
🔗 https://pretalx.seagl.org/2025/talk/9PWA3Q/

The Department of #Energy will host the #computers, the companies will provide the machines & capital spending, & both sides will share the #computing power, a DOE official said.

The 2 #supercomputers based on #AMD chips are intended to be the first of many of these types of partnerships with private industry & #DOE labs across the country, the official said.

#science #law #EnvironmentalLaw #AI #tech #electricity #Climate #ClimateChange #ClimateCrisis

#ORNL Director Stephen Streiffer said the #Lux #supercomputer will deliver about 3X the #AI capacity of current supercomputers.

The second, more advanced computer called #Discovery will be based around AMD's MI430 series of AI chips that are tuned for high-performance #computing. This system will be designed by ORNL, #HewlettPackard & #AMD. Discovery is expected to be delivered in 2028 & be ready for operations in 2029.

#science #law #EnvironmentalLaw #tech #energy #electricity #ClimateChange

The plans call for the first #supercomputer called #Lux to be constructed & come online within the next 6 months. It will be based around AMD's MI355X #ArtificialIntelligence chips, & the design will also include central processors & networking chips made by #AMD. The system is co-developed by AMD, #HewlettPackard Enterprise, #Oracle Cloud Infrastructure & Oak Ridge National Laboratory (#ORNL).

#science #law #EnvironmentalLaw #AI #tech #energy #electricity #Climate #ClimateChange #ClimateCrisis

"My hope is in the next 5 or 8 years, we will turn most cancers, many of which today are ultimate death sentences, into manageable conditions," Wright said.

[that sounds great, but #Trump has defunded #CleanEnergy, he actually wants to use #coal & other #FossilFuels which will increase #cancer rates!]

#science #law #EnvironmentalLaw #AI #tech #energy #electricity #Climate #ClimateChange #ClimateCrisis

…"We're going to get just massively faster progress using the computation from these #AI systems that I believe will have practical pathways to harness #fusion #energy in the next 2 or 3 years."

Wright said the #supercomputers would also help manage the US arsenal of #NuclearWeapons [Skynet anyone?] & accelerate #drug discovery by simulating ways to treat #cancer down to the
molecular level.

#science #law #EnvironmentalLaw #tech #electricity #Climate #ClimateChange #ClimateCrisis

"But every once in a while, Claude breaks bad. It lies. It deceives. It develops weird obsessions. It makes threats and then carries them out. And the frustrating part—true of all LLMs—is that no one knows exactly why." @stevenlevy for Wired

https://www.wired.com/story/ai-black-box-interpretability-problem/

#AI #LLMs #MechanisticInterpretability

#Energy Secy Wright said the systems would "supercharge" advances in #NuclearPower & #fusion energy, #technologies for #defense & #NationalSecurity, & the development of #drugs.

Scientists & companies are trying to replicate fusion, the reaction that fuels the sun, by jamming light atoms in a plasma gas under intense heat & pressure to release massive amounts of energy.

#science #law #EnvironmentalLaw #AI #tech #electricity #Climate #ClimateChange #ClimateCrisis

The US is building the 2 machines to ensure the country has enough #supercomputers to run increasingly complex experiments that require harnessing enormous amounts of data-crunching capability. The machines can accelerate the process of making scientific discoveries in areas the US is focused on.

#science #law #EnvironmentalLaw #AI #tech #energy #electricity #Climate #ClimateChange #ClimateCrisis

Exclusive: #US Dept of #Energy forms $1B #supercomputer & #AI partnership with #AMD

The US has formed a $1 billion partnership with Advanced Micro Devices to construct 2 supercomputers that will tackle large scientific problems ranging from #NuclearPower to #cancer treatments to #NationalSecurity, Energy Secy Chris Wright & AMD CEO Lisa Su told Reuters.

#science #law #EnvironmentalLaw #electricity #Climate #ClimateChange #ClimateCrisis
https://www.reuters.com/business/energy/us-department-energy-forms-1-billion-supercomputer-ai-partnership-with-amd-2025-10-27/?utm_source=braze&utm_medium=notifications&utm_campaign=2025_engagement

#Cats of #Halloween by Kaye Menner Wide variety #Prints & lovely #Products at:
https://kaye-menner.pixels.com/featured/cats-of-halloween-by-kaye-menner-kaye-menner.html
#autumn #pumpkins #digitalart #ai #mastoart #blackcats #fediverse #fediart #giftideas #wallartforsale #Art #artforsale #BuyIntoArt #AYearForArt #Artist #FineArtAmerica #PhotographyFeed #VisualArts #CreativeArts

Alt...

Cats of Halloween - My digital art rendition of 4 spooky black cats at Halloween looking after the property and minding the pumpkins. Digital art image enhanced in Photoshop. https://kaye-menner.pixels.com/featured/cats-of-halloween-by-kaye-menner-kaye-menner.html THE FINE ART AMERICA LOGO WILL NOT APPEAR ON PURCHASED PRINTS OR PRODUCTS.

#Guess has released an “#AI” video about its origins. We see the “AI” style and itʼs an instant turn-off, with its uncanny valley imagery: it has a distracting, artificial feel. What do you think?

https://www.instagram.com/reel/DQMhdpqjEiS/

For those of you thinking about AI use in its various forms in social scientific research, Due Cuong Nguyen & Catherine Welch suggest that this use is not only distorting analysis but actually leading researchers to make erroneous conclusions.

The paper is available open access; like all academic article its not an easy read, but if you're interested in AI (especially in university settings) then its likely of interest.

#AI #university #research

https://journals.sagepub.com/doi/10.1177/10944281251377154

Alt...

Conclusion We have shown that LLM chatbots are autoregressive models that generate word strings based on probability and ‘learned’ representations from their training data. These models excel at generating coherent synthetic text by predicting the next likely word based on statistical methods. As a result, while generated text may seem plausible, it lacks meaning, nuance and accuracy. Applied to qualitative data analysis, this leads to superficial outputs that imitate the form of established methodological practices and reporting templates—such as coding, theme generation and data structure tables—but that are neither grounded in nor justified by data. The superficial appearance of codes and themes not only creates a false sense of validity and depth but also conceals the absence of analytical engagement, while distracting and distorting the researcher's ability to uncover meaningful insights.

😂

#AI #bees

https://chaos.social/@gsuberland/115433648598393714