Free Harvard courses. Free Courses from Stanford. Free Courses from MIT. Free courses from Yale. Free courses from Princeton.

Free courses on Coursera.

Free Courses on EDx Free Courses on Alison

For paid, there's The Great Courses+/Wonderium. 20$ a month for unlimited courses.

And I also wanted to mention that humanitarian organisations have free courses too, like the Red Cross on international humanitarian law.

No transphobes allowed, only transborbs.

Check out my stuff!

It also goes without saying that bioessentialism inherently can’t be trans inclusive no matter how hard you try. “All men including trans men—” probably not. “This is only a woman’s issue—” is it really? “Afabs only—” why? “All trans men are like—” what? what are they like? finish the sentence i dare you.

Me watching idiots on Twitter/X whinging about how they want AO3 to be monetized and to censor all the ick ew omg stuff they hate and my special favorite: OMG WHY CAN'T THESE WEIRDOS GO OFF AND CREATE THEIR OWN SITE

::aid whispers in ear::

What? People have tried that before? Huh. Why hasn't--

::aid whispers in ear::

Not sometimes. Not until the next update. Always. They proved it with math.

Even with perfect training data and unlimited computing power, AI models will still confidently tell you things that are completely false. This isn't a bug they're working on. It's baked into how these systems work at a fundamental level.

And their own numbers are brutal. OpenAI's o1 reasoning model hallucinates 16% of the time. Their newer o3 model? 33%. Their newest o4-mini? 48%. Nearly half of what their most recent model tells you could be fabricated. The "smarter" models are actually getting worse at telling the truth.

Here's why it can't be fixed. Language models work by predicting the next word based on probability. When they hit something uncertain, they don't pause. They don't flag it. They guess. And they guess with complete confidence, because that's exactly what they were trained to do.

The researchers looked at the 10 biggest AI benchmarks used to measure how good these models are. 9 out of 10 give the same score for saying "I don't know" as for giving a completely wrong answer: zero points. The entire testing system literally punishes honesty and rewards guessing.

So the AI learned the optimal strategy: always guess. Never admit uncertainty. Sound confident even when you're making it up.

OpenAI's proposed fix? Have ChatGPT say "I don't know" when it's unsure. Their own math shows this would mean roughly 30% of your questions get no answer. Imagine asking ChatGPT something three times out of ten and getting "I'm not confident enough to respond." Users would leave overnight. So the fix exists, but it would kill the product.

This isn't just OpenAI's problem. DeepMind and Tsinghua University independently reached the same conclusion. Three of the world's top AI labs, working separately, all agree: this is permanent.

Not sometimes. Not until the next update. Always. They proved it with math.

Even with perfect training data and unlimited computing power, AI models will still confidently tell you things that are completely false. This isn't a bug they're working on. It's baked into how these systems work at a fundamental level.

And their own numbers are brutal. OpenAI's o1 reasoning model hallucinates 16% of the time. Their newer o3 model? 33%. Their newest o4-mini? 48%. Nearly half of what their most recent model tells you could be fabricated. The "smarter" models are actually getting worse at telling the truth.

Here's why it can't be fixed. Language models work by predicting the next word based on probability. When they hit something uncertain, they don't pause. They don't flag it. They guess. And they guess with complete confidence, because that's exactly what they were trained to do.

The researchers looked at the 10 biggest AI benchmarks used to measure how good these models are. 9 out of 10 give the same score for saying "I don't know" as for giving a completely wrong answer: zero points. The entire testing system literally punishes honesty and rewards guessing.

So the AI learned the optimal strategy: always guess. Never admit uncertainty. Sound confident even when you're making it up.

OpenAI's proposed fix? Have ChatGPT say "I don't know" when it's unsure. Their own math shows this would mean roughly 30% of your questions get no answer. Imagine asking ChatGPT something three times out of ten and getting "I'm not confident enough to respond." Users would leave overnight. So the fix exists, but it would kill the product.

This isn't just OpenAI's problem. DeepMind and Tsinghua University independently reached the same conclusion. Three of the world's top AI labs, working separately, all agree: this is permanent.

i technically had the choice not to update it, but if i didn’t, my insurance company would not cover the cost of fixing or replacing it if something happened to it. they could say that the damage came from not updating the system, even if the damage was completely external.

some of the software updates have been beneficial, and have made my life easier. however, the choice to use these new system has always been made for me. i can’t afford to replace this $3000 device that allows me to live without injecting insulin upwards of five times a day.

disabled people already live in a world that is hypermanaged by people who have absolute control over our health and safety. anyone with a prosthetic, hearing aid, or pacemaker can face the same problem i did if the manufacturer decides they need to exert their presence in our lives more than they already do. the “cyberpunk dystopia” is already here, and if you want to change the ever-growing vice grip of “smart technology,” help disabled activists. we’ve been here all along.

Digital files from my "Good News" zine

This work is about the joy and freedom that come with deconverting from an oppressive religion. I hope you enjoy!