A potential actually-useful use-case for chatbots: Back in The Good Old Days when programmers did not fear being swamped by bug reports, especially the not-so-great reports, software would carry a global-scoped 'gripe' button, when clicked, a redacted snapshot of the machine state was bundled with whatever they wished to put into the text box, and no reply was expected beyond 'received'. Given the ease of recognizing pure noise in the input, an LLM might coallate a bunch and produce a useful report? False positives happen in support all the time 😅

That kinda puts the #LLM in the 'triangle' position on a #StaffordBeer #VSM diagram, what which measures expected vs actual like a steam-engine governor, reducing Variety to useful patterns and all. 🤔

#cybernetics #unaccountabilityMachine #staffordBeer

Reading Dan Davies’ *The unaccountability machine*, a phrase popped into my head:

The Black Box event horizon: that point where a system grows so complicated the only thing to do is to shove all the complexity inside a black box and treat it as a series of inputs and outputs.

@henryfarrell

"The purpose of a system is what it does."

#StaffordBeer, 1926-2002

We live in an economic system that routinely gives more resources to people who make weapons than to doctors and nurses. What does that tell you about the purpose of that system?

#SystemsThinking #DeathDrive