以下部分引用 AI 总结，现阶段 AI 仍然有幻觉。请以内容的原文为准。

数据新鲜度群聊 07-14 ✓卖方 07-14 ✓Wrap 断档15天 ✕News 07-15 ✓X 断档12天 ✕页面生成 07-15

@gfodor

5 atoms · 跨 3 天 · 首见 2026-05-15 · 最近 2026-06-12

三色: 🟦 fact 0 · 🟥 take 5 · stance ▲1/▼4/◆0

来源: X 5

时态: fresh:5

标签: 好观点:4 · 好思考:1

2026-06-12 · 有东西要在AWE展示
God I have so many awesome things to show at AWE. If you want to see some magic, DMs are open!
tags: 好思考 · → daily
value: date=2026-06-12

2026-05-17 · argues token-level tax is absurd because models have wildly different costs per token
a token-level tax is totally absurd, because models all have wildly different costs per token to produce those tokens, by design.
tags: 好观点 · → daily
2026-05-17 · illustrates absurdity: tiny neural network taxed same rate as frontier models
If I throw up a tiny neural network that produces largely gibberish I am going to be taxed per token at the same level as the frontier models? Make it make sense!
tags: 好观点 · → daily
2026-05-15 · argues token-level tax makes no sense due to non-fungibility
this makes zero sense to me, given that tokens are perhaps the most non-fungible unit of economic value one could come up with in the entire field of AI.
tags: 好观点 · → daily
2026-05-15 · highlights absurdity of taxing a phone AI model same as Anthropic's Mythos
imagine taxing a company who is serving an AI model that can run on a phone the same as Anthropic's Mythos.
tags: 好观点 · → daily

2026-06-12 · narrative · 有东西要在AWE展示 · → daily
2026-05-17 · refute · argues token-level tax is absurd because models have wildly · → daily
2026-05-17 · refute · illustrates absurdity: tiny neural network taxed same rate a · → daily
2026-05-15 · refute · argues token-level tax makes no sense due to non-fungibility · → daily
2026-05-15 · refute · highlights absurdity of taxing a phone AI model same as Anth · → daily