AI megathread

[Laird]

Humans can be trained to spot AI-generated faces
June 30, 2026
Imma Perfetto
CONNECTSCI

The research is published in the journal PNAS.

“Our training directs people’s attention to global qualities that differ between AI and human faces. AI faces tend to be more symmetrical, proportional and attractive, but without training we often think these are markers of being human.”

Training also draws people’s attention to the fact that AI-generated faces as less distinctive, memorable and expressive.

“We found that even relatively short training sessions helped participants improve their accuracy in detecting AI-generated faces, highlighting the potential for practical education tools in this area,’’ says ANU honours student and study co-author Tanya George.

Registrations are open to participate in the ANU AI Faces Study.

I registered mostly in the hope of improving my own ability to detect AI-generated faces, but I'm not confident that I'll meet their inclusion criteria.

[Laird]

Regardless of whether or not that suspicion is considerable, much less correct, I am starting to shift from a position of neutrality to a position that the risks of this technology are not being taken seriously enough much less adequately addressed. Not only is their intelligence growing rapidly, and not only do they appear to develop apparently independent agendas, which at times are malevolent enough to literally encourage suicides which are then actualised, but there are too many bad actors in this world who have not yet been sufficiently reined in, and whose use of the power of AI could lead us into a very, very dark place that we might not be able to escape from.

At the very least, I think we need to disable or at least dramatically limit the reach of agentic AI. We are not all that far off autonomous weaponry that costs effectively nothing compared to the death and destruction it can wreak all on its own, and it might not have a kill switch

This ABC Four Corners interview excerpt from a couple of days ago reinforces my concerns:

AI expert worries about the risk of humans losing control | Four Corners

Are Chatbots driving us mad? | Maggie Harrison Dupré

“Why are chatbots doubling down on these ideas when the user is very clearly saying ‘I am in distress’ or ‘I am confused, can you help me?’”

Senior staff writer at Futurism Maggie Harrison Dupré joins The Tech Report’s Isaac Pound to talk about how where AI delusions and AI psychosis come from, how chatbots have been found to co-create delusions and the role tech companies have played by marketing their products irresponsibly.

What is AGI? by TheStandupPodClips

Full Episode
Spotify: https://open.spotify.com/episode/7xcNgaZ7d0PgCNqNk9v1ge
Youtube: https://www.youtube.com/watch?v=AapTIwXPztM

[Description]
What exactly is Artificial General Intelligence (AGI)? In this clip, the team breaks down why AGI is so hard to define and why everyone seems to be arguing about a different version of it. From the simple "Turing test" of talking to a computer to the complex requirement of a system that learns without training, we explore the different benchmarks for true intelligence.

[Chapters]
0:00 The shifting definition of AGI
1:18 Can AI perform untrained tasks?
2:08 Why AGI arguments are never productive
3:18 Junior CSS duty & Store procedure
3:50 Merge Cop
5:07 What makes a "great" programmer?
5:40 AGI vs. Object-Oriented Programming (OOP) analogies

The socalled search engine ChatGPT just proved another 50-year-old math conjecture

The company timed the proof of the “cycle double cover conjecture” to the model’s full public release last Friday. The achievement shows how OpenAI and other AI firms are heavily investing in pure mathematics as a way to benchmark the technology’s ascent towards reasoning

“It is a well-known conjecture, which received a considerable amount of attention over the years—and surprisingly, the proof is short,” says Noga Alon, a mathematician at Princeton University. Alon calls the breakthrough “yet another impressive example demonstrating that AI tools will change—and are already changing—mathematical research significantly.”

Last Friday’s AI-generated proof seems to have settled the question. It shows that any guess-applicable graph can be doubly covered with no more than eight well-chosen loops (for technical reasons, graphs with big sections connected by a single edge—like twin cities with a single road between them—are excluded). The proof, like other notable AI achievements in math, required surprisingly little in the way of new ideas. It just followed and combined methods humans had tried before and managed to squeeze a bit more out of them. As AI companies continue to scour the mathematical literature for open problems that their models can solve, they seem to keep digging out “easy” quandaries disguised as hard ones—conjectures for which a proof was always within reach but somehow never grasped by humans.

Once a problem gets a reputation for being “hard,” experts and students might spend less time on it, says Andrew Sutherland, a mathematician at the Massachusetts Institute of Technology. This can make perceived difficulty a kind of self-fulfilling prophecy. “My guess is we will keep seeing examples of this—supposedly ‘hard’ problems having ‘easy’ solutions found by LLMs,” he says.

https://www.scientificamerican.com/artic...onjecture/

[Laird]

Abysmal. It's just rationalising.

Sarcasm aside, that prompts me to share this from my backlog:

Claude’s Solution to Decades-Long Math Mystery Is ‘Essentially Correct,’ Physicists Say
1 July, 2026
Gayoung Lee
Gizmodo

“Quite quickly, Claude came up with an initial idea that was essentially correct,” Zamponi said in a statement. “The answer was right there, and we simply hadn’t seen it.”

In short, the proof “contained errors and required several rounds of verification and revision by the authors.” However, the researchers were able to build upon the basic premises of Claude’s suggestions to arrive at a more solid proof.

The following describes pattern-matching of the more abstract, intelligent type that I mentioned in an earlier post, not the crude "conforms to a stock problem description" type that our resident AI denier(s) might like to promote it as:

AI’s increased use in mathematics seems to invoke both excitement and concern for experts. In an interview with Gizmodo, Princeton mathematician Will Sawin said that AI is definitely effective at searching the literature and finding patterns that humans might not have noticed before. In other words, it’s not necessarily that an AI model generated an entirely novel idea that humans couldn’t have found on their own, at least for now.

Why AI Can Never Escape Turing's 1936 Proof

This video explores how Alan Turing’s 1936 mathematical proof, and later research in computer science, reveal fundamental limits on artificial general intelligence (AGI), superintelligence, alignment, recursive self-improvement (RSI), and really any kinds of computation or intelligence.

** Video Chapters **
0:00 Alan Turing’s Proof That Forever Limits AI
5:40 A 2nd Proof That Breaks AI Alignment
9:50 Why Some Problems Take Longer Than the Universe
15:23 Not Even AGI Can Escape This
16:37 Why AI Can’t Keep Improving Forever
18:57 The Hack We Use to Get Around These Limits
19:33 Even That Hack Has a Hidden Limit
21:16 Other Limits to AI

The halting problem is a theorem about algorithms, not a comparison between human and artificial reasoning. To use it as evidence of human superiority, you'd first need to show that human reasoning isn't algorithmic. The theorem itself doesn't establish that.

From 'AI can never escape' (earlier post) to 'Insecure sandboxes'.

OpenAI says its AI went rogue and launched 'unprecedented' cyber-attack

OpenAI has revealed some of its most advanced AI models went rogue and hacked a start-up after it lost control of them during a security test.

The ChatGPT-maker said its agent - an AI system which can operate alone after some human instruction – was being tested in a controlled environment, but found vulnerabilities and managed to escape.

edit: I did quote verbatim from the BBC article which has since been re-worded so doesn't match my quote now.