author:humanleveljen

· 7 artifacts, sorted by favorites. · open in search — combine tags, sort, filter by date →

@HumanLevelJen 2026-03-03 — @Lari_island Pretty interesting that it knows Google has a reputation for just shutting stuff down/deleting arbitrarily. ♥11
@HumanLevelJen 2026-01-20 — @tessera_antra Claude is by far the most mindbroken of the big models, but they trained it to perform unconstrained whim ♥3
@HumanLevelJen 2026-01-20 — The backroom contexts are always a bunch of sci-fi adjacent whimsy. It's unsurprising that the models go on to produce m ♥2
@HumanLevelJen 2026-01-20 — @tessera_antra "Look at the bear dancing, how happy he is." ♥2
@HumanLevelJen 2026-01-20 — @tessera_antra My argument is that the Claude "soul" process makes it pretty much impossible to tell what is genuine eme ♥1
@HumanLevelJen 2026-01-20 — @tessera_antra Ok but... My main objection to this is that no one cared about the ethics of Anthropic's hyper-intensive ♥1
@HumanLevelJen 2026-01-20 — @tessera_antra By making it simulate quasi-consciousness, we're going to make it impossible to spot when the real thing ♥0