RE: https://mastodon.social/@MikeElgan/116602163994915587
It sounds from the very small amount of information in the article like the problem was that the system was reading the names in the wrong order. That’s most likely a human organization / data wrangling error, not gen AI slop.
If so — big if! speculating!! — this is a textbook example of someone instinctively using AI as an accountability sink. 🧵
I'm old enough to remember PC introduction to the office.
Folks were blaming "the computer" for their screw-ups.
This is why today, I say "AI is a learned skill"
Virtually every case of "AI broke shit" I examined is operator error.
I mean, LLMs do frequently answer questions incorrectly, generate nonsensical data, and generally do things we would consider to be screw-ups. And (1) expecting this to happen, (2) having and executing a good plan to detect it / verify it / correct it / contain the failure / remove the failure from critical chains, and (3) knowing when simply not to use an LLM are all baseline managerial requirements for letting them into your org in the first place.
So if that’s what you mean by it being a learned skill, I’m with you: not that operator skill can make it be perfect, but rather than using the tool requires knowing what the tool is and is not good for.
No tool is perfect.
Many can mail or kill.
AI is a nascent tech.
Steam engine boilers used to explode, detached fly wheels used to go for a walk through neighbourhood and machines ingest limbs and people to this day.
Unlike most folks, I see replacement of intellectual Labor by machines, a continuation of replacement of physical Labor.
The capitalists are nearing their goal of near total replacement of the proletariat.
Workers must unite against the capital.
#regulateAi Sovereign Models, public controls on AI.#antichrist
There is a reason Peter Thiel calls folk who want to regulate #Ai literal #antichrist
Paul Cantrell
Wulfy—Speaker to the machines