

We’ll see how that works out when nobody creates new answers to new questions. Ten years from now LLMs have nothing but hallucinated trash for training data. How do you troubleshoot anything at that point is a mystery.
VGhlcmUgaXMgbm8gZ2VudWluZSBpbnRlbGxpZ2VuY2UgLCB0aGVyZSBpcyBhcnRpZmljaWFsIHN0dXBpZGl0eS4NClRoZXJlIGlzIG5vIHNlcmVuaXR5LCB0aGVyZSBpcyBhbnhpZXR5Lg0KVGhlcmUgaXMgbm8gcGVhY2UsIHRoZXJlIGlzIHR1cm1vaWwuDQpUaGVyZSBpcyBubyBzdHJ1Y3R1cmUsIHRoZXJlIGlzIHBvcnJpZGdlLg0KVGhlcmUgaXMgbm8gb3JkZXIsIHRoZXJlIGlzIGNoYW9zLg==


We’ll see how that works out when nobody creates new answers to new questions. Ten years from now LLMs have nothing but hallucinated trash for training data. How do you troubleshoot anything at that point is a mystery.


Yeah, that’s the vibe in getting in these conversations. Nobody seems to have any issues with image sorting, optical character recognition, audio signal processing or smart background blurring.


AI is a very broad field. LLMs are just one corner of it. Probably the most visible one.


A completely unrelated idea just popped in my head. There’s a reason why the Russian army is targeting the Ukrainian power grid.
What about an aboleth or a tarrasque? Either one would be pretty cool.