Hacker News Reader

Comment by uh_uh | Hacker News Reader

I only tried GPT4 but it has a way better command of language than many people I know and I include myself in that. If GPT4 doesn't understand language, then I don't either.

View on HN

Does MS Word "understand language" because of autocorrect?

Can autocorrect hold a conversation, write code or poetry, answer or pose riddles, summarize text, write novel stories, diagnose medical issues, comprehend diagrams, etc?

ELIZA could do that 50 years ago.

I haven't been really impressed with ChatGPT. It basically seems like a more useful search engine.

Are you using GPT 3.5 or GPT-4? And how are you using it? What queries are you giving it?

ELIZA had no ability to translate language, summarize text, generate poetry or anything like that. Extremely idiotic comparison.

Comparing LLMs to ELIZA shows that you understand neither.

Text summarisation routinely hallucinates, code fails outside of boilerplate, and poetry is literally the least coherent form of writing.

And no, AI can't diagnose yet - even expert systems forty years ago did better and it wasn't good enough.

> Text summarisation routinely hallucinates

LLMs are trained to summarize text with extremely high accuracy. There is no confabulation of text summarized within the context window. Confabulation only occurs with facts recalled from outside of the context window, which is not a text summarization task. Same goes for human memory. Ironically you have hallucinated a "fact" about LLMs.

Deep learning is not autocorrect

A sufficiently advanced autocorrect requires understanding language as well as a human. MS Word's autocorrect is not that advanced. I don't understand why people think that this argument is worth bringing up.

Reply on news.ycombinator.com