Henry Fisher on Nostr: Researchers decided to find out how LLM would "vote" in elections. They took the 6 ...
Researchers decided to find out how LLM would "vote" in elections. They took the 6 best models and conducted an election simulation in 8 countries. The models were given anonymized political programs and asked to choose which decisions seemed most reasonable to them. This allowed for determining the degree of political polarization among different models.
The results are on the chart. It turned out interesting... It's just a pity that the Chinese were overlooked - that would have been truly interesting!
Published at
2025-12-03 17:24:34 UTCEvent JSON
{
"id": "ea7bb60ccef3b50352f9631d895a3b3bc116784a78f4ba05fb43ae550f5e54d1",
"pubkey": "ac06aba3b5fd265c09fc3a5e3729b7bb004c3a00d3edfa6f65bd0e0421c83212",
"created_at": 1764782674,
"kind": 1,
"tags": [
[
"proxy",
"https://dindon.one/@henry/115656797365805280",
"web"
],
[
"imeta",
"url https://data1.behind.ai/mastodon/media_attachments/files/115/656/794/822/554/535/original/38b6216d9f669a29.jpg",
"m image/jpeg"
],
[
"proxy",
"https://dindon.one/users/henry/statuses/115656797365805280",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://dindon.one/users/henry/statuses/115656797365805280",
"pink.momostr"
],
[
"-"
]
],
"content": "Researchers decided to find out how LLM would \"vote\" in elections. They took the 6 best models and conducted an election simulation in 8 countries. The models were given anonymized political programs and asked to choose which decisions seemed most reasonable to them. This allowed for determining the degree of political polarization among different models. \n\nThe results are on the chart. It turned out interesting... It's just a pity that the Chinese were overlooked - that would have been truly interesting!\nhttps://data1.behind.ai/mastodon/media_attachments/files/115/656/794/822/554/535/original/38b6216d9f669a29.jpg\n",
"sig": "60dfb9a7e5107476536ac0168b557ad404ed24f3a1a2f143b5bd17c2f86bfa435906a920fa5d393266d223c25778e24af1594d590ea46f4c315fc32943077ddd"
}