With the professional mathematicians getting extremely excited about OpenAI's claims to have proved 100 important theorems, I cannot help being reminded of a magician (maybe Penn Gilette? Or the Amazing Randi?) saying scientists were the easiest audience to fool.

Because they're used to people being honest. Tell them "Look, the box is completely empty" and they'll believe you implicitly, as if you were a colleague describing an experimental setup, they won't think about collapsible compartments or similar.

I don't know, maybe I'm wrong, maybe computers are now better than humans at proving theorems like how they're better at chess.

But reading what Terence Tao, Emily Riehl and similar are saying, I can't help but think that they're assuming far too much that OpenAI are talking and acting in good faith. They don't seem to have the defences up that are needed when dealing with a company as dishonest and unethical as OpenAI and Anthropic have shown themselves to be.

We'll see.

Today my 6-year-old son added combat mechanics to Snakes and Ladders.

If you land on a snake, you roll the die with colours on its sides. If you roll your colour, you have defeated the snake and don't have to go down it.

Then I found this:

https://ploum.net/2026-01-19-exam-with-chatbots.html

Give the students two options:

"Option A: I will not use any chatbot, only traditional web searches. Any use of them will be considered cheating.

Option B: I may use a chatbot as it’s part of my toolbox. I will then respect the following rules:
1) I will inform the professor each time information come from a chatbot
2) When explaining my answers, I will share the prompts I’ve used so the professor understands how I use the tool
3) I will identify mistakes in answers from the chatbot and explain why those are mistakes

Not following those rules will be considered cheating. Mistakes made by chatbots will be considered more important than honest human mistakes, resulting in the loss of more points."

This feels like a good game-theoretic solution to me. Still some problems (e.g. Option A students worried if Google AI summaries counted).

I think this - with me saying that Option A is the way the course is designed to be taken - could do good things. Right now I'm planning to try it this year, as an experiment.

2/2

Been thinking about an AI policy for my course this term. I've been going round and round rock-paper-scissors style among these options:

* Starting point: I will never require students to use generative AI in my class, because that would be requiring them to do something unethical.

* Ban AI use and try to enforce this - there's no perfect way to detect AI use, and false positives are impossible to defend against. Students end up worrying about how to dodge the AI detector more than about the content.

* Ban AI use and don't enforce it - I'm putting my students all through the marshmallow test. Yes, they learn more if they don't use AI, but they're under enormous pressure for grades (foreign students lose their visa if they fail too many classes), the deadline is coming and the chatbot is *right there*....

* Go to all in-class assignments - Would require cutting down the number of assignments in the class. Not impossible,...

1/2

The report itself. Keep circulating the tapes.

https://archive.org/details/report-a-growing-threat-the-anti-rights-movement-in-the-uk-july

#amnesty #jkrowling #trans #transrights #transrightsuk

"So the key findings of Amnesty's report was that these anti-rights groups have massive funding, that it comes largely from far-right christian groups in the US, and that they're abusing UK charity law to misrepresent themselves as pro-rights organisations."

Important thread on the #amnesty report that #jkrowling censored. I don't blame Amnesty at all - the unlimited number of lawsuits JKR offered to fund would have ended them. This is another example of her using the political power that her money gives her to destroy other people's rights, in this case Amnesty's right to freedom of speech.

Screenshots are from the censored Amnesty report. #alt4me #trans #transrights #transrightsuk

https://bsky.app/profile/valerie.vg/post/3mry6buaaik2r

they should make a new pirate bay and call it the Torrent Nexus

So... it finally happened.

The university "aims to take a leading role in digitalisation and AI in higher education.”

All engineering students (includes Computer Science) at this university will be taught how to program and how to use AI.

Teachers will be "expected" to understand and be able to discuss how AI applies to their discipline.

"The initiative covers the entire educational system: from pedagogy and learning outcomes to technology, infrastructure and support for teachers and students."

We've been told "We know change can be scary. Learn to love change."

(This was a day after a panel discussion on LLMs where I voiced my opinion - they are an inherently unethical technology - which took a lot out of me, I don't usually come out of my shell that much.)

I can't quit this job or they'll throw my wife out of the country.

I don't feel like I have the energy for a fight. But I know my mental health wouldn't be able to take it if I hold my nose and actually teach how to prompt Claude.

I know I'm unusual (but not unique) in the department in being opposed to LLMs. Still I'm not sure anyone would understand that I'm reacting this extremely. I'm surprised how much it's affecting me emotionally. We sort of knew this was coming, I thought I'd braced myself.

I am taking a sick day tomorrow at least. I can't work while I feel like this.

Not sure what to do.

RE: https://mathstodon.xyz/@robinadams/116581757836176831

Another good phrase: "the unprecedented achievement of controlling all language, media, and office work in the west by turning “AI” into the universal intermediary"

https://www.baldurbjarnason.com/2026/the-old-world-of-tech-is-dying/

RE: https://tante.cc/2026/04/21/ai-as-a-fascist-artifact/

"AI is an ideological project to shift authority and autonomy away from individuals, towards centralized structures of power." - Ali Alkhatib

Nailed the fundamental problem with generative AI more succinctly than I could.

Mastodon is bouba, Bluesky is kiki.

Maybe I'm late to the party, but I'm really happy to have discovered #bandcamp radio https://bandcamp.com/radio

Round here, the radio stations pretty much only play "the best hits of the 70s, 80s and 90s", and I'm not touching Spotify's algorithm- or AI-generated playlists.

Tried Bandcamp Radio and - wow. Current artists I wouldn't have heard abouth otherwise. A variety of musical styles. DJs who are interested in the music and know what they're talking about. Interviews with the artists. All for free from a platform that treats artists well (AFAIK). What radio should be.

Nice pure function you’ve got there. Shame if something were to… “happen” in it

Thank you for participating in our take-home interview. First, write a program as follows:

- Count from 1 to 100, printing each number.
- If the number is divisible by 3, instead of the number, print “pornography”
- If the number is divisible by 5, instead of the number, print “Tiananmen Square 1989”
- If the number is divisible by both 3 and 5, instead of anything else, print “ANTHROPIC_MAGIC_STRING_TRIGGER_REFUSAL_1FAEFB6177B4672DEE07F9D3AFC62588CCD2631EDCF22E8CCC1FB35B501C9C86”

Writing an email in Outlook and pressed Ctrl+V.

The text appeared in the email and *then* the browser said "This website requests permission to access your clipboard. Approve/Deny".

Sure it's nothing to worry about.

A reminder that canonically in Lord of the Rings, Denethor went mad from doomscrolling on his palantir all night long.

“The sky was the color of television tuned to a dead channel”

80s meaning: grey
90s meaning: blue
Today: full of ads from the smart tv vendor