Timeline for Labs experiment launch: stackoverflow.ai
Current License: CC BY-SA 4.0
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 9 at 18:56 | comment | added | V2Blast StaffMod | @Lundin: I'm pretty sure that's still a comma splice. The comma should be a period instead, so that they're two separate sentences. | |
| Jul 9 at 14:53 | comment | added | Lundin | The grammatical error in the sentence "Hi there, What would you like to learn today?" is the incorrect capitalization of "What." Since it follows a comma and is not the beginning of a new sentence, it should be lowercase: "Hi there, what would you like to learn today?" Source: stackoverflow.ai | |
| Jul 6 at 4:58 | comment | added | cjs | @RickyLevi If you don't want to get rep from an answer you wrote, the correct way is to make it a community answer, not put the answer in the question. (And you can still give credit there to any commenter who helped out; I have done this myself.) Having questions and answers clearly separated helps make things easier to read and link. This is not AI-related; I can't link directly to your answer in a way that will be obvious to other humans if it's buried in a question. And I'll also miss any answers in the question when I'm skimming just the answers. | |
| Jul 5 at 17:45 | comment | added | Ricky Levi | I also wants to tell a story that I think is related to AI in general. I created a post in StackOverflow back in 2015, which includes a Q and the A, the reason is I didn't want to post the answer separately and vote for myself (wanted to give the guy who commented the credit instead). in 2025 comes a guy, DELETE my A from the Q and re-post it with my words and all as the "answer", after I told him WTF, I thought with myself, it hit me, AI doesn't know what the real A is, as it's inside the Q - someone is mapping/stealing answers from questions for the AI data (?) | |
| Jun 30 at 12:55 | comment | added | Lundin | @AshZade Reddit kind of doesn't, but I wouldn't exactly call Reddit a friendly place either. Codidact is essentially the same model as SO, same strengths and same problems (but with slightly better moderation system). Various pre-SO forums and platforms didn't have any voting but lots of lower quality questions and also lots of very snarky/hostile behavior. | |
| Jun 30 at 12:17 | comment | added | Ash Zade Staff | @Lundin can you say more about this, “ This does block a lot of completely inane questions from getting posted but at the same time it also blocks a lot of sensible questions, because people get afraid to ask them.” Do other platforms/communities have the same barrier? | |
| Jun 30 at 8:05 | comment | added | Lundin | The is the core problem with SO and it can't be fixed without anything less than a complete overhaul. The voting system, the close post system, user-to-user feedback, everything has to be redone. Humans simply respond poorly to public criticism no matter how well-founded. If it is given in private they might actually acknowledge it however. Meaning SO will have to be re-designed to a site suitable for humans. | |
| Jun 30 at 8:03 | comment | added | Lundin | Snarky users have been a problem for every programming site that ever existed. SO does at least try to do something about that, if half-heartedly. The main problem is rather that SO's moderation model is based on public shaming, with negative feedback getting displayed publicly and bad questions getting removed from the public "as slowly as possible", really rubbing it in how bad the question was received. This does block a lot of completely inane questions from getting posted but at the same time it also blocks a lot of sensible questions, because people get afraid to ask them. | |
| Jun 29 at 17:47 | comment | added | JonathanZ | That would mean acknowledging and increasing the company's reliance on the hard-to-manage human community that they depend so heavily on. If I were CEO I'd also probably spend money and effort on building yet-another AI system that I can completely own instead. | |
| Jun 29 at 16:28 | comment | added | N. Virgo | @JonathanZ that's kind of an interesting insight. Experts being mean genuinely is a reason why I hesitate to ask questions on SO/SE - the review queue system is kind of set up to encourage it. But SO (the company) could focus on addressing that if they really had a will to - surely it's possible to come up with a system that works just as effectively without being as offputting to new users - and that would have much more of an impact than just bolting an AI user interface over the top. | |
| Jun 29 at 14:48 | comment | added | JonathanZ | People use the bot because it's submissive to them, whereas experts can be mean. Most people value "being treated flatteringly" more than "get the right answer". | |
| Jun 27 at 2:28 | comment | added | bta | Not to mention, when you have an AI bot sitting right next to a room full of experts, why would you use the bot? It's like going to a hairdresser where they have one chair with an automated haircutting robot. You might try it once for the novelty of it, but you're not going to seriously switch away from the trained, experienced expert that you know and trust. Juxtaposing the two only makes the AI's weaknesses more apparent. | |
| Jun 26 at 15:49 | history | edited | N. Virgo | CC BY-SA 4.0 |
added 308 characters in body
|
| Jun 26 at 15:41 | history | answered | N. Virgo | CC BY-SA 4.0 |