I don't know if it's possible or what the timelines are, and neither does anyone else. So I don't tend to get worked up about it, unless someone is trying to control me with p(doom) scare stories.

What follows is an audio and textual record of a talk on AI given by Jon Stokes and Julie Frederickson at The Wagon Box on June 28, 2025.

A detailed look at recent papers on the limits of reasoning in LLMs.

Claude and I drifted apart, and it turns out it was my fault.

From Claude Shannon in the late 1940's to DeepSeek R1 and beyond.

Sometimes cheating is good, actually

Of roguelikes and grounding problems

3D flight simulators and Turing tests are cool, but some of us have real work to do

This week’s post is a guest essay by my friend Greg Fodor, a former colleague, brilliant programmer, and most recently the brains behind shogtongue. Greg has been reading, hacking, and tweeting in the AI space for some time now, and while he’s not an…

AI is both extremely powerful and entirely new to the human experience. Its power means that we are definitely going to make rules about it, and its novelty means those rules will initially be of the “fighting the last war” variety and we will mostly…

1 Comment <div pseudo="-webkit-input-placeholder" id="placeholder" style="display: block !important;">Write a comment…</div><div></div> © 2023 Jon Stokes Privacy ∙ Terms ∙ Collection notice Start WritingGet the app Substack is the home for great…

Housekeeping note: I’m still working on the follow-up to my most recent post on RLHF — basically a close look at the preference model that acts as a proxy for the tastes and morals of a group of model humans. But instead of pushing that out, today, I…

There was a page made about fixing AI with journalism recently noting the market opportunity for mainstream news to regain the public's trust, along with other ways to aid the news media like making local journalism more efficient and with "pair…

The story so far: In the previous installment of this series, I described RLHF as the fine-tuning phase where we endow the ML model with a moral compass or a sense of what is good and bad. But this “moral compass” talk, which I’m essentially doubling…

The story so far: We’ve all read the endless commentary on ChatGPT’s political biases, and in fact, I’ve written a few tweets on this topic, myself. But where do these biases come from? How is this large language model, trained on trillions of words…

I read with interest a new piece on AI risks by RETURN contributor Jon Askonas and economist Samuel Hammond. It’s a sensible counter to some of the AI doomer alarmism that’s currently fashionable, and I endorse its general approach to the big…

No, there's many problems that this article ignores. "Micropayments/sender-pays" was one of the first and obvious ideas for fighting spam, and it wasn't due to The Feds that it failed. The problem with "micropayments" is that it's a hand-wave, and…

An excellent example of why this straightforward bias paradigm fails, and is kind of obvious nonsense reflecting deep and embarrassing thinking errors on the part of the perpetrators, is to look at something like the "pound of feathers" trick. Many…

Note to subscribers: Apologies for the lack of updates so far this month. The whole family (including me) caught mono in the last half of April, and then we got strep and colds on top of that. So that was big fun! But I’ve recovered now and am…

2 Comments <div pseudo="-webkit-input-placeholder" id="placeholder" style="display: block !important;">Write a comment…</div><div></div> Dave Reed53 min agoLiked by Jon StokesSounds like we need to beat some folks with our lamps. 🪔 The whole thing…