123

Effective immediately, the moderators of Academia.SE (wrzlprmft, cag51, and Bryan Krause) are on strike. This is part of the network-wide action described here, and follows an Academia moderator resignation a few days ago. We will not perform any moderation functions until this situation is resolved. Other users are welcome to join us by refusing to perform moderation functions such as voting, voting to close and editing posts (but everyone will make their own choice; please don't harass users who continue to participate normally).

Why is this happening?

We moderators have been ordered not to intervene when AI-generated content is posted to our site, except under very limited circumstances. While the exact details are not public, even the publicly-available guidance effectively allows almost all AI-generated content site-wide, and admits that "this standard would exclude most suspensions issued to date." In other words: our policy banning AI-generated answers is unenforceable; automatically-generated content cannot be treated differently than human-generated content.

It is hardly necessary to explain why this is a problem generally (and others already have). But here on Academia.SE, we are particularly concerned because many of our users are asking for advice on life-altering career decisions. In such a setting, automatically-generated content is not a mere nuisance, but can cause irreparable damage to someone's life. While we recognize that AI-generated content is increasingly difficult to detect, and there will be false positives and false negatives, the current policy (don't do anything at all) is the worst possible response.

Is this an overreaction?

Not really, this seems to be our only option short of allowing AI-generated content to run rampant over our network. Moderators network-wide have been pushing back against this policy for a week already to no avail. On the contrary, we've been told about another policy in the pipeline that will make the situation even worse. And the company has historically shown little flexibility even when they were clearly wrong**; as a result company-moderator trust is not very high.

When will this end?

Our sole demand is that any network-wide policies must allow this stack to delete the majority of automatically-generated content and message/suspend users who post it. We acknowledge that there will be false positives and false negatives, and so we are willing to take guidance. But a policy that effectively allows all automatically-generated content is unacceptable.

Alternatively, this will end when we all get fired and new moderators are elected. That's fine, we don't get paid (even the swag we were promised never arrived), and there is little point in sticking around so we can impotently preside over a hellscape where bots talk to bots. In that case....we appreciate you for having elected us; so long, and thanks for all the fish.

** For context: in 2019, a popular moderator was fired and slandered publicly; after a huge uproar, the company issued a lawyer-speak apology but refused to reinstate the moderator. Instead, their solution was to form a "moderator council" to make us feel heard, but the council never had much influence and has since disbanded. Even so, the individual community managers have been great, and we had successfully built a certain amount of trust before this incident destroyed everything again.

9
  • 15
    One of the StackOverflow moderators posted a good "how did we get here" post over on their Meta: meta.stackoverflow.com/q/424979/7233542
    – Bryan Krause Mod
    Jun 5, 2023 at 2:15
  • 10
    Jon Ericson, former SE moderator and former SE employee (CM), also has a series of blog posts commenting on the situation at jlericson.com
    – Bryan Krause Mod
    Jun 5, 2023 at 2:19
  • 8
    Very well written, thanks. Let me also add that here we've been particularly careful in not suspending unnecessarily for AI-generated content, and that we suspended only in blatantly abusive cases. Jun 5, 2023 at 5:56
  • 50
    @Philippe, why did you remove the featured tag?
    – henning
    Jun 5, 2023 at 12:32
  • 12
    @henning I guess either to support the strike and help get the word out or in a misguided effort because they are embarrassed.
    – Bryan Krause Mod
    Jun 5, 2023 at 12:59
  • 9
    @henning Unless he had talked with the mod team about it, he is violating his own policies by doing so.
    – forest
    Jun 6, 2023 at 10:00
  • @cag51 The company did not actually issue an apology, in the sense of admitting that the termination was unjustified.
    – einpoklum
    Jun 7, 2023 at 21:12
  • 2
    Update 7 June: No time for a summary now, but Phillippe posted some data (from StackOverflow) that they used when crafting this policy here. It is an interesting read. I wrote an Academia.SE-specific response here.
    – cag51 Mod
    Jun 8, 2023 at 1:24
  • Wow the edit is depressingly similar to my experience as a reddit mod for one of the biggest subreddits around the same time. They made their own councils after really stepping in it with the redesign, breaking a lot of features of reddit, after already tarnishing the relationship with mods by not supporting them with keeping the site sane. Now they're in the midst of their own backlash. It's very interesting watching two of the most popular websites undergoing similar turmoil. Jun 12, 2023 at 4:54

5 Answers 5

39

Thanks Cag. I hope nobody will begrudge my answering to associate myself with these remarks and the strike. This isn't really a problem solely for moderators, though I very much appreciate the mods being so quick to inform the stack at large about the ongoing extent of SE's misconduct. Certainly I don't see a reason to keep filtering content for them if they intend to fight actively against our ability to do that in a responsible way. If they want to be like Yahoo Answers, that's still going strong, right (/s)?

1
23

I think this is the only valid response to such abysmal decisions. One thing I wonder, is it possible to post this message (or a shorter one) to the main page? I feel quite many (newer) users don't even know meta exists and it would be great to inform them, too.

3
  • 7
    This is featured, so it’s visible (on the desktop site at least) from the main page. I don’t think it would be appropriate to post meta content there directly (though if you did, the mods would not stop you at the moment…). If you mean a banner or something, that’s not something we can do ourselves.
    – cag51 Mod
    Jun 5, 2023 at 2:46
  • 41
    @cag51 It's no longer featured. SE staff is actively unfeaturing meta posts about the strike.
    – Mast
    Jun 5, 2023 at 9:55
  • Although it's not featured, it is under "hot meta posts".
    – gerrit
    Jun 13, 2023 at 7:37
12

I wonder what SE sees as the worthwhile benefits to users of posting AI generated stuff here ?

But maybe it's just a slick way of "tuning" AI generators so the output can "fool" real people.

I'm 100% behind the strike I almost didn't know about.

12
  • 6
    I don't think they want AI-generated stuff here. Rather, they seem to be convinced that a ton of humans have been wrongfully accused of being Chat-GPT. It is sometimes tricky to tell the difference, and mistakes are inevitable....but this is the worst possible response. Frankly, if a human is writing stuff that is indistinguishable from ChatGPT, then it doesn't belong here anyway. We require answers to be based on references or personal experience in academia...we don't check ID on the latter, but it's often easy to tell.
    – cag51 Mod
    Jun 6, 2023 at 7:07
  • 3
    My personal assumption is that someone in the SE management went along the following lines of thought: "More users => more content => more profit" - and noticed (or perhaps even only thought) that users creating "content" being suspended... failing to notice that said "content" was actually AI-generated - i.e. possibly good looking at first glance, but highly likely to be inaccurate at best, complete garbage at worst.
    – CharonX
    Jun 6, 2023 at 9:21
  • 1
    What ? You are saying the SE Academia forum is being scamped with AI generated answers ? I haven't seen any though as a non-mod scanning limited postings I wouldn't expect to. I read the odd post that affected a novelistic idea of academia but that was spotted by everyone else. This wasn't AI - just someone playing around.
    – Trunk
    Jun 6, 2023 at 9:56
  • 1
    @Trunk There are a handful here, some of them as spam seeds. StackOverflow gets thousands.
    – Bryan Krause Mod
    Jun 6, 2023 at 14:12
  • 1
    I think it looks awkward for them to have an official anti-AI policy while they're trying to pivot to AI. What they want is moderators to suspend users posting AI-generated content for low-quality content without explicitly saying that it's being banned for being AI-generated. Jun 6, 2023 at 14:51
  • @NoahSnyder It's perfectly possible for them to pivot to AI while still being against AI-generated content intermingling on the site. All of the AI uses SE has suggested so far are compatible. It would be better for keeping content here useful for future AI training, and is necessary to keep these sites distinct from places where someone can simply ask an AI prompt directly. Of course those issues may not be salient to the person/people making these decisions.
    – Bryan Krause Mod
    Jun 6, 2023 at 17:19
  • You wonder, because they don't bother to tell us. And they don't bother to tell us, because they have no obligation to tell us (and because they don't think that our collaboration, or opinions, or responses matter). And that's part of the problem.
    – einpoklum
    Jun 7, 2023 at 21:15
  • @CharonX is spot on. It's all just Google Food. If Google likes it, that translates into more $$$ for SE. TBH, the users are irrelevant, except when they're looking at ads or generating content. If the machines can generate content, all the users are good for is ad targets. Jun 7, 2023 at 23:22
  • 1
    Well, how many are up for a new independent Academia Forum then ? Maybe some hosting company would offer free hosting, the forum software isn't too dear and I suppose people can maybe live without the job adverts tickers . . .
    – Trunk
    Jun 7, 2023 at 23:30
  • 2
    I'd recommend Codidact; they seem to already have OS SW similar to SE. No Academia site yet but there is a process for proposing new sites. But I personally will probably go down with this ship...after answering questions here for five years, I don't have much more to say, if it comes to leaving, I will probably move onto other endeavors rather than finding a new platform.
    – cag51 Mod
    Jun 8, 2023 at 20:52
  • Hello @Trunk, I agree with cag51: Codidact seems to be the best place (even though they, definitely, don't know how to pick nice names for sites (just a joke from me)). If you're going to propose Academia on Codidact, I'll sign up. Jun 10, 2023 at 21:31
  • @Ivan Nepomnyashchikh Lovely name you've got there, Ivan. But as an only occasional contributor with strong views against tenure (if you're good enough to be given it, you don't need it; if you can't do your job without it, you don't deserve it), against university-paid (rather than students' union-paid) counselors and not least the fact that I am no longer in academia, I would not be a popular choice as founder member.
    – Trunk
    Jun 11, 2023 at 10:42
7

Much respect and full support to @cag51, to @wrzlprmft and to @BryanKrause.

A strike song for you :-)

It's preposterous for SE Inc. to enact this policy - and this just illustrates how the mechanism of them dictating policies arbitrary is itself broken and inappropriate.

Please don't hesitate to ask me - or users in general - for concrete measures of support to the extent that you need them. And yes, that definitely includes off-site action, and even monetary support if there are legal expenses or if someone needs to take time off their day job etc.

I personally believe the strike's demands should be expanded and deepened, as I elaborated in my MSE post regarding the strike; but like I said - full support for your current demand.

-5

Sorry, what was this all about?

SE has this page from 8 months ago https://meta.stackoverflow.com/questions/421831/temporary-policy-generative-ai-e-g-chatgpt-is-banned?cb=1

where they state that AI-assisted content generation is not allowed. Are they rewriting history or simply playing with the moderator?

4
  • 1
    What do you mean by rewriting history? A number of individual sites discussed and implemented generative-AI policies after the launch of chatGPT late in 2022. Late May 2023 is when Stack Exchange (the company) imposed a new policy (without disclosing crucial details to the userbase at large), effectively overruling the previous policies. The strike was in response to this policy.
    – Anyon
    Aug 29, 2023 at 13:55
  • 1
    Since you went back on Meta to 8 months ago, perhaps you could look around the time that this was originally posted in late May and after - SE went back on their pronouncement and changed the rules, including implementing secret mod-only rules.
    – Jon Custer
    Aug 29, 2023 at 14:47
  • 3
    As others have said, the whole point of the strike was because SE was unilaterally disallowing us from enforcing policies like the one you found. Their new policy (secret at the time, but now public) was that AI-generated comment could only be removed if the author freely admitted that it was AI generated. The strike succeeded and policies like the ones you found are still in force.
    – cag51 Mod
    Aug 29, 2023 at 16:22
  • @JonCuster I have like 30secs to dedicate to browsing random stuff. The Meta I posted is one that was "promoted" on the top right box "Featured on Meta". I think this content is curated, isn't it? I fortunately have no interest whatsoever in running all the crap SE publishes on Meta. I still wonder how much time can one dedicate to help such malignant entity, though.
    – EarlGrey
    Aug 30, 2023 at 8:58

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .