Why Nostr? What is Njump?
2024-11-26 04:49:31

Darth Hideout 🏳️‍🌈 on Nostr: When I was wee, computers were the size of barns & yet there were still vending ...

When I was wee, computers were the size of barns & yet there were still vending machines. Before bill scanners & card readers, these mechanical contrivances could achieve something humans require intelligence, training & practice for: They could sell merchandise & even make change. They mainly distinguished direct physical characteristics of coins such as weight, diameter, thickness, presence of milling, etc. But this was enough for a machine to do a job that required a human to think.

At no time did any such device ever think.

The same goes for every #computer program ever written. Computer programs, like vending machines, are state machines, cleverly contrived (at most) to respond to different real-time conditions. We often lazily say that a computer makes a decision, eg based on your input, it “decides” what to do. This is false. The human who wrote the program made the decision. The program is responding to its input as it was designed, by a human, to do. It’s never the machine doing the thinking or deciding. That human process of thought & design may be faulty & error-prone; the operation of the machine may surprise the mechanic. This doesn’t mean the machine is thinking. It means the mechanic is not thinking enough.

And of course, the same goes for modern #AI. The machines are far more elaborate and contrived to use (not “understand”) natural language; to obtain (not “discover”) source material on the internet; to acquire state (not “learn”) to extend the capabilities offered. Lazy terminology is common & inspires fantasies.

Honestly, I watch my Roomba & I use the same words to myself: It “can’t figure out” how to get around a certain corner or it “decided” to take another pass down the hallway. It’s hard to avoid, but it doesn’t thereby become true.
Author Public Key
npub1k532ws9032mgmtt7zufzq0k73lgwqnu2apzzwm39lj8dgh6kcvpspctcfw