Timeline for Was the new AI search feature "test" officially announced?
Current License: CC BY-SA 4.0
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 6 at 8:54 | comment | added | Shadow Wizard | @YaakovEllis well, I did move on, but what can I do when there is a fresh AI scandal? | |
| Dec 6 at 7:12 | comment | added | Yaakov Ellis | @ShadowWizard "but it really points on some very fundamental problem". Lay off please. You are preaching based on partial knowledge. This stuff happens everywhere. Just more visible here. And things are done to prevent this from happening. But the fact that it still happens is a sign that there are humans involved. You made your point. Move on. | |
| Oct 2 at 21:00 | comment | added | Shadow Wizard | @CesarM that is the vibe I get here, as time after time this happens, and it feels like nothing is done to prevent the next "accident". | |
| Oct 2 at 15:21 | comment | added | Cesar M StaffMod | @ShadowWizard I don't think it's fair to characterize it as "Meh it's fine, only an accident". I've never said that. Accidents can be quite serious; there's nothing about saying it's an accident that diminishes its seriousness. Rather, accident talks about intention: i.e.; the pilot didn't intend to throw the airplane into the sea; it was an accident. There's a lot that can be learned from accidents and tons of safety measures to be applied from these learnings. That doesn't make the accidents less accident of an accident. | |
| Oct 1 at 20:25 | comment | added | wizzwizz4 | @ShadowWizard The fundamental problem is with the whole world's attitude towards computers, not with Stack Exchange in particular. If only we'd listened to Dijkstra, none of this would've happened… | |
| Oct 1 at 18:13 | comment | added | Shadow Wizard | @CesarM sorry, but that was glaring obvious to see, and it's not the first, not the second, and not the third (and we can keep counting) time this happens, i.e. something that isn't ready yet somehow finds it way to Meta, or the whole network. You're only the messenger, but it really points on some very fundamental problem. (I think the "meh it's fine, only an accident" approach is contributing for this, and means it will just happen again, with some other "feature" that should not be seen or released yet.) | |
| Oct 1 at 17:56 | comment | added | Cesar M StaffMod | @CodyGray not to say that there isn't an issue/gap (there have been a few cases recently, it's true), but we do have code review. | |
| Oct 1 at 17:18 | comment | added | Cody Gray | I think there may be a gap in y'all's process for pushing/enabling things on production. Specifically, I think you may be missing the part where a second developer needs to confirm before something can get enabled on the public platform. This is very critical, because anyone can make a mistake. And, unfortunately, we're seeing a lot of mistakes recently. I don't want to blame anyone for a mistake, because that's just not productive. But it is a real problem writ large, and the fix for that is often getting a double-check inserted somewhere in the process, where 2 devs have to sign off. | |
| Oct 1 at 14:47 | comment | added | AMtwo | @ShadowWizard Feature Flag shenanigans, most likely. | |
| Oct 1 at 14:41 | comment | added | Cesar M StaffMod | @ShadowWizard No details of implementation that I can share, no. But yeah, just an accident :) | |
| Oct 1 at 14:33 | comment | added | Shadow Wizard | Any details how this "accidental turning on of the feature" happened, technically? Such thing requires: 1. change in codebase, 2. push to production, 3. approval and deployment, 4. ? | |
| Oct 1 at 14:30 | comment | added | Starship | Thank you. From what I saw of the AI search it was making up complete nonsense and misinformation. | |
| Oct 1 at 14:28 | vote | accept | ꓢPArcheon | ||
| Oct 1 at 14:24 | history | answered | Cesar MStaffMod | CC BY-SA 4.0 |