How is that a useful analogy? Its not like a person at all it just wears the mask of one. Its still the same architecture underneath. Still a computer running calculations and not a human thinking.
Nonsense. People already usefully anthropomorphize their technology. I am finding the analogy much more helpful as that technology does more anthropomorphic tasks.
“Before it starts working on the bug, refresh its memory about the coding standards and design docs.”
“It gets confused when the context window is too full of unrelated info – too much in its head all at once.”
“It does a lot better with these small, easy-to-understand examples.”
“It does tend to hyperfocus on one solution, following down a blind alley. Reminding it of this propensity helps it to recognize when it’s doing that.”
EDIT: It’s all bitwise NAND operations at the hardware level. You’re fooling yourself using these “high-level” abstractions… like C, Python, …
Which usually works, but anthropomorphizing AI is dangerous as we have seen. It leads people to believe it thinks, which it does not. Its highly misleading for nontechnical people and I believe a large part of why AI psychosis happens so frequently now than it did the last time chatbots were popular.
How is that a useful analogy? Its not like a person at all it just wears the mask of one. Its still the same architecture underneath. Still a computer running calculations and not a human thinking.
Nonsense. People already usefully anthropomorphize their technology. I am finding the analogy much more helpful as that technology does more anthropomorphic tasks.
“Before it starts working on the bug, refresh its memory about the coding standards and design docs.”
“It gets confused when the context window is too full of unrelated info – too much in its head all at once.”
“It does a lot better with these small, easy-to-understand examples.”
“It does tend to hyperfocus on one solution, following down a blind alley. Reminding it of this propensity helps it to recognize when it’s doing that.”
EDIT: It’s all bitwise NAND operations at the hardware level. You’re fooling yourself using these “high-level” abstractions… like C, Python, …
Which usually works, but anthropomorphizing AI is dangerous as we have seen. It leads people to believe it thinks, which it does not. Its highly misleading for nontechnical people and I believe a large part of why AI psychosis happens so frequently now than it did the last time chatbots were popular.