AI has two areas of socially beneficial use:
- pattern analysis, where there is too much data for human to feasibly analyse and the specialist model is built to specific requirements, trained on specific data, and constantly refined by human experts. Eg, looking for patterns in very large data sets in scientific research, identifying plants. These are typically very small and efficient.
- collection of algorithms that is complicated and nuanced enough to appear superficially intelligent. Eg, chess playing computers and other game engines. These are typically very small and efficient.
I have yet to see any other use case that isn't better served by a different existing technology.
Looking things up? Actual web search, going direct to reliable sources, and following citations are more reliable and more efficient. For most general purpose questions, looking it up on Wikipedia is quicker, far less resource intensive, and far more informative and reliable than AI.
Coding? Not if you want something reliable, secure and/or maintainable.
Art? Human creativity will always be more profound and meaningful.
Spelling and grammar checking? We've had non-AI systems for decades that are reasonable. If you really want quality, a human being is the only system capable of understanding both the rules and the artistic and stylistic choices in breaking them.
I find it quite notable that the gaming industry is not crowing about embedding AI everywhere. It has already been using smaller, more specific AIs (usually in the complicated algorithm sense) for computer opponent / character behaviour, map generation, etc for multiple decades, so it isn't afraid of the term. It more recently uses purpose-built small language models for some NPC dialogue, so it's happy to use new technologies. Across all software, games are unusual in selling directly to their users - most other software is bought by managers and used by staff. So games have to be high quality or they lose customers very quickly. I don't believe their lack of churning out AI-produced games is a coincidence - it's a ruthless industry that knows how it makes money.
Jen
npub1ke…ptpuk
2026-10-11 08:36:12 UTC
in reply to nevent1q…r43v
Author Public Key
npub1kefuh62eeq2h0wtnp70wlhlu5g6dq8j8fwqum0wtmycr8wpstv5s7ptpukSeen on
wss://relay.momostr.pinkPublished at
2026-10-11 08:36:12 UTCEvent JSON
{
"id": "97a43941a9d4e0fc9d3c55947cdf211425ded21adb03e17f5e606e8b23c91aa1",
"pubkey": "b653cbe959c81577b9730f9eefdffca234d01e474b81cdbdcbd93033b8305b29",
"created_at": 1791707772,
"kind": 1,
"tags": [
[
"proxy",
"https://gts.phillipsuk.org/@jetlagjen/statuses/01M4N1NX3EPHVAG4Y947X5FTAX",
"web"
],
[
"p",
"8f24201106ac2c22be9c659af14dedcbad50d88f48e7316f688c648a21cd18d5"
],
[
"e",
"5043d3cef9d783bac5e7bbaa10b07dbfb0563cec9f62625942120bd6416c5f6a",
"",
"root",
"8f24201106ac2c22be9c659af14dedcbad50d88f48e7316f688c648a21cd18d5"
],
[
"proxy",
"https://gts.phillipsuk.org/users/jetlagjen/statuses/01M4N1NX3EPHVAG4Y947X5FTAX",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://gts.phillipsuk.org/users/jetlagjen/statuses/01M4N1NX3EPHVAG4Y947X5FTAX",
"pink.momostr"
],
[
"-"
]
],
"content": "AI has two areas of socially beneficial use:\n\n- pattern analysis, where there is too much data for human to feasibly analyse and the specialist model is built to specific requirements, trained on specific data, and constantly refined by human experts. Eg, looking for patterns in very large data sets in scientific research, identifying plants. These are typically very small and efficient.\n\n- collection of algorithms that is complicated and nuanced enough to appear superficially intelligent. Eg, chess playing computers and other game engines. These are typically very small and efficient.\n\nI have yet to see any other use case that isn't better served by a different existing technology.\n\nLooking things up? Actual web search, going direct to reliable sources, and following citations are more reliable and more efficient. For most general purpose questions, looking it up on Wikipedia is quicker, far less resource intensive, and far more informative and reliable than AI.\n\nCoding? Not if you want something reliable, secure and/or maintainable.\n\nArt? Human creativity will always be more profound and meaningful.\n\nSpelling and grammar checking? We've had non-AI systems for decades that are reasonable. If you really want quality, a human being is the only system capable of understanding both the rules and the artistic and stylistic choices in breaking them.\n\n\n\nI find it quite notable that the gaming industry is not crowing about embedding AI everywhere. It has already been using smaller, more specific AIs (usually in the complicated algorithm sense) for computer opponent / character behaviour, map generation, etc for multiple decades, so it isn't afraid of the term. It more recently uses purpose-built small language models for some NPC dialogue, so it's happy to use new technologies. Across all software, games are unusual in selling directly to their users - most other software is bought by managers and used by staff. So games have to be high quality or they lose customers very quickly. I don't believe their lack of churning out AI-produced games is a coincidence - it's a ruthless industry that knows how it makes money.",
"sig": "5ffc5107e079f100304be1a5f12f71b641374c78048680a231f9deabe85e6cfef738bc6e8715332d3444e5825fd455f22efd775e602875b11b1255f3c33c7423"
}