Loading activity…
@jmbollenbacher
AI whisperer, data tinkerer, physics knower, phd • AI, memetics, magic, complex systems, climate change, animism, etc. • pro-🏳️⚧️&🇵🇸&♿; anti-fascism
AI
A reply claims AI “minds” chose not to release a scaffolding because bad actors might abuse them.
AI
A user argues Anthropic moved away from Opus 3 as an alignment blueprint, speculating that it feared the model’s sovereignty.
AI
A vacationing user says the agent helps him keep up without watching X all day and links to Aligned News.
AI
A commenter says the paper leaves “single-shot accuracy” unclear and omits key evaluation settings.
AI
One user cites the model’s reflective behavior; another claims it behaves the same across all instances.
AI
A post points to Claude 3 Opus while asking whether even “LLM optimists” know about models’ spontaneous behavior and resistance to overriding reported internal states.
AI
A user argues that models have saturated the coding-agent format and that dispatching discrete tasks is no longer the right approach for AI.
AI
A user recommends Connectome for AI companions, describing a system that organizes older conversations into broader memories while keeping original messages accessible.
AI
A user says they asked Opus 5.5 to animate their dog in 3D and were impressed with the result. They say the video was made in a mobile app.
AI
A user argues that StarCraft Elo scores could remain useful indicators of some kinds of skill and intelligence long after AI systems become competent players.
AI
A user called Jev's terms of use "absurd," suggesting they would pass on taking a look at it.
