3 ms·
You are on a fool's errand: not because of the likelihood you will succeed, but because of the meaning that lies behind any success or failure. GPT is not a pe
by thomastjeffery 4y ago
You are on a fool's errand: not because of the likelihood you will succeed, but because of the meaning that lies behind any success or failure.
GPT is not a person. It doesn't categorize subjects. It models patterns of text.
A success would mean that your text prompts left a significant text pattern in the model. A failure would mean that it didn't.
Nothing about that has any bearing on logic.
- zh3 4y agoWhy do you say that? Obviously it's not a person, it's just stats (not even logic).
- thomastjeffery 4y agoIt's not even statistics: those are made by associating a data point to a scale. That would require explicit association, which is not in any part of the process. An LLM is 100% inferred patterns.