5 ms·

No offense meant to parent commenter if I'm wrong, but looking at their comment history, does anyone else suspect this is a GPT-3 bot? I've been playing a lot i

No offense meant to parent commenter if I'm wrong, but looking at their comment history, does anyone else suspect this is a GPT-3 bot? I've been playing a lot in the GPT Playground this last week and this is very similar to what I get there.

ivanmaeder 4y ago
The bot's recommendations are good, although maybe not inspiring. "Programming Pearls" might be an exception—there are some great real-life stories and problems to think through in that. Maybe also "The Art of Computer Programming" but I can't speak from experience (I'm not ready to tackle that one yet).
- yesenadam 4y ago
  Gee, I've read 6 or 7 and found all of them extremely inspiring. (I've not read Code or CLRS Algorithms, and only parts of SICP and TAoCP) Well ok, 'inspiring' maybe is not exactly the word for Man-Month, but it is fascinating. Looks like the list of 10 most-recommended programming books, and things don't get on that list for no reason!
aristofun 4y ago
agree, the comment feels off the point, while not being off topic
- Jtsummers 4y ago
  It reads like a copy/paste from a listicle. However Google (no big surprise these days) failed to turn up results for me with the exact phrasing.
Phithagoras 4y ago
The phrase > here's our list of 10 of the best. seems particularly listicle-y, especially with the full stop rather than a colon or nothing
- AnimalMuppet 4y ago
  To me, the thing that jumps out is the "our". That's not the normal way an HN comment would be written. If it's not a bot, it's someone speaking for a group - a publication or a company. (Of course, a GPT3 bot could have lots of that kind of writing in its training corpus.)
want2takearide 4y ago
It does feel like a GPT-3 bot. I tried testing the GPT3-Davinci model with similar prompts and I get 3 to 5 of the same 10 recommendations.
yesenadam 4y ago
> looking at their comment history, does anyone else suspect this is a GPT-3 bot Yes, I've never before on HN thought a commenter was non-human, but most of their comments begin with similar rehashing the topic in a very weird way, like they can't stop writing ad copy. Like they were given the topic and are responding in magazine article or super-commercial-blog style - they don't have the flavour of a person talking a person. I don't think that should be allowed on HN.

ivanmaeder 4y ago

The bot's recommendations are good, although maybe not inspiring. "Programming Pearls" might be an exception—there are some great real-life stories and problems to think through in that. Maybe also "The Art of Computer Programming" but I can't speak from experience (I'm not ready to tackle that one yet).

yesenadam 4y ago
Gee, I've read 6 or 7 and found all of them extremely inspiring. (I've not read Code or CLRS Algorithms, and only parts of SICP and TAoCP) Well ok, 'inspiring' maybe is not exactly the word for Man-Month, but it is fascinating. Looks like the list of 10 most-recommended programming books, and things don't get on that list for no reason!

aristofun 4y ago

agree, the comment feels off the point, while not being off topic

Jtsummers 4y ago
It reads like a copy/paste from a listicle. However Google (no big surprise these days) failed to turn up results for me with the exact phrasing.

Phithagoras 4y ago

The phrase > here's our list of 10 of the best. seems particularly listicle-y, especially with the full stop rather than a colon or nothing

AnimalMuppet 4y ago
To me, the thing that jumps out is the "our". That's not the normal way an HN comment would be written. If it's not a bot, it's someone speaking for a group - a publication or a company. (Of course, a GPT3 bot could have lots of that kind of writing in its training corpus.)

want2takearide 4y ago

It does feel like a GPT-3 bot. I tried testing the GPT3-Davinci model with similar prompts and I get 3 to 5 of the same 10 recommendations.

yesenadam 4y ago

> looking at their comment history, does anyone else suspect this is a GPT-3 bot Yes, I've never before on HN thought a commenter was non-human, but most of their comments begin with similar rehashing the topic in a very weird way, like they can't stop writing ad copy. Like they were given the topic and are responding in magazine article or super-commercial-blog style - they don't have the flavour of a person talking a person. I don't think that should be allowed on HN.