Haha or colonoscopy.

Sounds more fun than a recipe site.

Is each image a NN or is it one NN for all the images?

One NN per pattern/image (instance based training).

The doers can't remove the bureaucracy-focused people, because doing so would run counter to Pournelle's Iron Law; it's like crossing the streams. Let's instead replace bureaucrats with AI, since the same pressures that drive bureaucracies toward self-preservation might instead push AI systems toward AGI .

The law doesn’t define behavior. It describes behaviors.

Those behaviors can change by giving the doers more power. If a bureaucrat is trouble, and the doers can cause the bureaucrat to be fired, to doers will have more power.

Wasn't able to follow install instructions for bonsai_term. Have you seen ncurses? It's really easy to install.

> I think it's interesting to see in which ways the models fail and that they actually produce some good stuff once in a while.

This is a good summary of GPTs.

Yes it happened to me yesterday with Deepseek. No one died who wasn't already dead.

Yikes don't ask Openclaw to fix this.

SQL injection and XSS come up in dynamic analysis too.