4 ms·
Other than 89 I’m not impressed with the GPT-3 output here. In fact I would say it’s a good demonstration of the limits of this technology. While syntactically
by WoodenChair 6y ago
Other than 89 I’m not impressed with the GPT-3 output here. In fact I would say it’s a good demonstration of the limits of this technology. While syntactically correct sentences that at the surface sound good, if you really think about them they’re not saying anything. They’re much more superficial and meaningless than the thinking that went into the original list.
- hnhg 6y agoI'd say that GPT-3 could probably beat a lot of the English-speaking (human) population at producing these though. I've read far worse on social media, that I've attributed to humans authors.
- WoodenChair 6y agoNo doubt. Since it’s built out of a corpus of human produced text, it should be capable of producing human level text (but probably not often exceeding it). Yet many people are not capable of writing syntactically correct text.
- hnhg 6y agoAs well as being syntactically correct, it also reflects the contextual patterns of the tokens used to train it and then prompt it, which is perhaps limited by both intelligence and knowledge in humans.
- AsyncAwait 6y agoSo GPT-3 could run for office.
- borroka 6y agoIt sounds like 95% of the conversations I hear at work.