Post #4168403
2026-07-28 14:37 UTC
#ai #llm #claudecode
There's no way I'm the first person to realize the context of LLMs/AI is _literally_ remote code execution on your device.
...
This is a thing we've been trying to prevent for _decades_. Everything we do is just awful.
Replies (1)
-
@401matthall@mastodon.xyz 2026-07-28 14:54
EVERY... SINGLE... PART... Of this training is about creating... DETERMINISM... Where it _CANNOT_ exist. The logical fallacy here is _SO_ wildly significant, the literal _crux_ of the entire discussion around _usefulness_ or _usability_. This is so wildly exhausting. This is expecting the results from decades of learning and practice in a given skill from a machine that is _incapable_ of providing consistent results. (I'm haven't even gotten to the theft and environmental violence.)