4 ms·
Tell HN: So, I Tried Bard
I tried Bard.
It worked. I tried it because ChatGPT could not have a sensible conversation with me about some code and we were talking in circles.
Bard solved it immediately. This is the result of my single try of Bard so not very scientific, but an encouraging start.
- pk-protect-ai 3y agoWould you show the example? I didn't invest a lot of time with Bard, just did my usual tests, and when found that it can't properly solve the traveling salesman problem (unlike GPT-3,GPT-4), I have rejected it as a possible tool.
- pk-protect-ai 3y agoTried the Bard again, just now... It is horrible. It is no way near the GPT-4 level, and it is just a little worse then gpt-3.5-turbo (ChatGPT free version). It hallucinates a lot more then gpt-3.5-turbo, it loops in it's own previous wrong and already corrected solutions. Nope. Not my tool ...