Posts tagged with ai
Agentic AI outside coding
The level of competence, adaptiveness, and general capability even qwen-code (known to be a pretty bad agent framework/app compared to the best ones) displays is genuinely insane. I can just give it a task, turn on a sandbox and YOLO mode, and it will execute exploratory find, grep, curl, ls, which etc commands to figure out the structure of its environment and the data I'm giving it, and then write and execute command line scripts (or python, and install dependencies, and use curl to access
Do AI images and video mean "reality is broken?"
A lot of people seem to be angry, upset, or panicking about the advent of generative AI image and especially video models like Veo 3 and Sora 2, claiming that since these systems allow us to effortly create photorealistic videos and images completely out of whole cloth, our society is finally completely screwed: we have no way of getting information that we can know comes from actual reality. We can no longer trust anything.
Leaving aside the possibility of technological solutions — such as
Explaining LLMs to laypeople
Introduction
This is an attempt to explain large language models (e.g. ChatGPT or DeepSeek) in such a way that a layperson could actually understand what's going on under the hood — not just in terms of lazy metaphors (whether overhyped or dismissive) but in terms of the actual algorithms they use.
I think this is really important, because if we don't understand what AI is, it's very easy to get misled, either by overestimating its abilities, using it for the wrong tasks, or underestimating
Large language models will never be able to reason reliably… but that doesn't make them not useful
The fundamental structure of large language models is not just different from the structure of the human brain, but different in a way that fundamentally leans them away from truth and reasoning capabilities. The differences include:
- Trained on a reward function that only optimizes for generation of superficially human-seeming (by probability) token sequences, not token sequences which correspond to true or useful ideas or accurate reasoning or anything of the sort. Human beings, on the oth
On Gary Marcus
I have Gary Marcus in my blogroll. I agree with his idea that neuro-symbolic architectures are the way forward for robust AI.
<div style="border: 1px solid white; padding: 5px; margin: 5px;">
Side note: Although unlike him:
- I do not think that causation is fundame
The phenomenology agentic coding
The discursive trap of productivity metrics
AI coding agents are important not because they make one more productive, but because they fundamentally alter what it is like to program. That is what interests me and — I believe — should interest you, about this whole enterprise. The productivity discourse is, to a large degree, misguided noise, a discursive trap motivated by reactionary Luddism on the one hand and market hype on the other. While AI coding agents certainly make me, as a pr
Why am I being so mean to indie artists? Am I a tech bro?
To be perfectly clear, the purpose of this post, and all my other posts on this page expressing frustration at popular views concerning information ownership and "intellectual property," is not to punch down at independent artists and progressive activists. I care a lot about them, because I'm one, and I know many others; I'm deeply sympathetic to their values and goals and their need for a livelihood.
The reason I write so much about this topic, directed as often if not moreso at independent