darren buck on Nostr: We should stop saying that machine-learning algorithms are 'trained.' They aren't ...
We should stop saying that machine-learning algorithms are 'trained.' They aren't capable of learning, of critical thought, of creativity. They aren't being 'trained.'
They operate on probabilities based on data fed into it. They hopefully spit out better answers with more data calibrating their models. That's not 'training.'
Published at
2023-09-18 13:48:07Event JSON
{
"id": "78246ef34f00669290081b07f27993e47ec2a49fb2a6f2ce1d73212159bcdf5d",
"pubkey": "e02d7e67f27e544d18f17a1197ae7bbc2934e4b365236a75dc27f96d07e2bbe4",
"created_at": 1695044887,
"kind": 1,
"tags": [
[
"proxy",
"https://transportation.social/users/bikepedantic/statuses/111086461744654961",
"activitypub"
]
],
"content": "We should stop saying that machine-learning algorithms are 'trained.' They aren't capable of learning, of critical thought, of creativity. They aren't being 'trained.' \n\nThey operate on probabilities based on data fed into it. They hopefully spit out better answers with more data calibrating their models. That's not 'training.'",
"sig": "273a5793db832a13fab35fdd1ade2bcb7b83fd4a4c8938d949875bc721828877e9c364cddf53087083d13c259cd71825fe29802eeb3ecfced94b03cb7fe59c8c"
}
