AI and Bots

[Armchair Cyclist]

This post, and the next 11, were moved from another thread: the OP for this thread is actually post 13, but this previous discussion pre-emps it and seemed to me worth moving.

Your help please, for a moderator who is aware of, but doesn't really get, this AI stuff.

My apologies to them if my suspicions are misplaced, but the posts by josephews and now Noah147 seem somehow too inane and anodyne to seem quite human. {The rest of you can take that as a compliment: you are sufficiently objectionable for me to accept your humanity unquestioningly}

Joseph has posted some links, but none that are obviously spamming, so I'm not sure what to make of them: what is the point of anyone sending bots here? (again: great apologies to Joe and Noah if you are breathing sentient beings). Is the intention that they will return and edit the posts to fill them with spam links (some have been here a while without that happening: there were some links in Joe's posts in the Summer Holiday thread, now deleted, but that has not been typical)?



I have avoided replying directly or quoting them, which might trigger a notification, which is why I haven't @ atted them here, but even then, what would they stand to gain? They would know that whoever replied is active here, but unless they have burrowed into the software, they are not getting any info on us.

I am bemused, and perhaps a little naive: your thoughts please.

(I reject typically about 12 new account requests a day that the forum software throws up as suspect, and I presume that the software intercepts a lot more before it reaches my approval list)

 

[SHAD0W93]

Is AI able to correctly @ someone though? Otherwise both look sketchy.

 

[hrotha]

Imagine being so bland that you're mistaken for a bot at cycling news dot com, there's no coming back from a burn like that

(But yes they don't pass the smell test)

 

[Nick2413]

There could be humans pressing the actual button to post but the content itself is definitely bot driven

 

[Armchair Cyclist]

But why? I don't get the bot motive.

And is it harmful?

 

[Tim Cahill]

But why? I don't get the bot motive.

And is it harmful?

The current meta is to mostly post comments that can be construed as normal, usually generic comments about films, locations etc to fool people into thinking it's a real person in order to then astroturf and put forward their ideas. They will have a database of millions of forums using a bot to create accounts on, mostly it's done to push forward political ideas or advertise a product, this one seems to be a Czechia tourism bot.

It's harmful on a mass-political scale, can theoretically be used to sway tight elections, but becoming less effective as time goes on and people start to recognise the signs. On this kind of forum anything like this will usually be benign, usually something to push cycling/exercise gear.

There are also leaked password databases scanning billions of accounts so they can take over old accounts on every website you can think of, because usernames with lots of posts have more credibility. If you go into your old outlook email and check attempted logins you will see hundreds of attempts at accessing it from every VPN location you can think of.

 

[AmRacer]

The current meta is to mostly post comments that can be construed as normal, usually generic comments about films, locations etc to fool people into thinking it's a real person in order to then astroturf and put forward their ideas. They will have a database of millions of forums using a bot to create accounts on, mostly it's done to push forward political ideas or advertise a product, this one seems to be a Czechia tourism bot.

It's harmful on a mass-political scale, can theoretically be used to sway tight elections, but becoming less effective as time goes on and people start to recognise the signs. On this kind of forum anything like this will usually be benign, usually something to push cycling/exercise gear.

There are also leaked password databases scanning billions of accounts so they can take over old accounts on every website you can think of, because usernames with lots of posts have more credibility. If you go into your old outlook email and check attempted logins you will see hundreds of attempts at accessing it from every VPN location you can think of.

If there's a link in the post, it can be potential harmful.

 

[Netserk]

Thanks for the heads up on those flicks, @Unchained! We actually dodged a bullet with Blink Twice last weekend. Almost watched it, but went with Brightburn instead. But before that, we tried giving afrAId a shot, but decided it's not worth finishing.

Hehe, notice the uppercase letters. #Straussian

 

[topcat]

Imagine being so bland that you're mistaken for a bot at cycling news dot com, there's no coming back from a burn like that

(But yes they don't pass the smell test)

I can think of a few.

 

[rallybanana]

There could be humans pressing the actual button to post but the content itself is definitely bot driven

My favourite bot was the one who would copy old posts but change some of the words to make it look like original content. The giveaway was when it changed Lance Armstrong's name to Spear.

 

[AmRacer]

My favourite bot was the one who would copy old posts but change some of the words to make it look like original content. The giveaway was when it changed Lance Armstrong's name to Spear.

Spear Legweak

 

[RedheadDane]

My favourite bot was the one who would copy old posts but change some of the words to make it look like original content. The giveaway was when it changed Lance Armstrong's name to Spear.

All those "strings"...

 

[Armchair Cyclist]

This is the OP of this as a specific thread, but I will be moving some earlier discussion into this from other threads, so it may not be the first for long

There has been a rapid growth, but still a fairly low number, of posts that seem to come from bots or to be AI generated.

I have asked Admins, and their approach seems to be that so long as there is not spam-links involved, they are unconcerned, and no rules are being broken. Meanwhile, I am getting many people reporting them, presumably in an expectation of intervention. I have removed those that are way off topic (badly programmed AI?), but left others.

I don't know (maybe others here are more alert to the capabilities) whether these are monitoring whether they get responses and likes in the forum. If they are, then emoticon responses and quotes are only going to encourage them, so please don't.

They are not necessarily breaking any rules, but I believe that a common tactic is to have these programmed to come back and edit posts with links that would normally be picked up in new post.

I am also concerned that there could be a volume of posts that becomes disruptive: we have gone from a couple a week to about three a day very rapidly, and that rate of growth continued could swamp discussion. I wonder whether we might need to re-introduce a scenario that (IIRC) previously held whereby any new member had to have their first few posts approved before they became publicly visible.


Thoughts/insights?

 

[Potomac]

This is the OP of this as a specific thread, but I will be moving some earlier discussion into this from other threads, so it may not be the first for long

There has been a rapid growth, but still a fairly low number, of posts that seem to come from bots or to be AI generated.

I have asked Admins, and their approach seems to be that so long as there is not spam-links involved, they are unconcerned, and no rules are being broken. Meanwhile, I am getting many people reporting them, presumably in an expectation of intervention. I have removed those that are way off topic (badly programmed AI?), but left others.

I don't know (maybe others here are more alert to the capabilities) whether these are monitoring whether they get responses and likes in the forum. If they are, then emoticon responses and quotes are only going to encourage them, so please don't.

They are not necessarily breaking any rules, but I believe that a common tactic is to have these programmed to come back and edit posts with links that would normally be picked up in new post.

I am also concerned that there could be a volume of posts that becomes disruptive: we have gone from a couple a week to about three a day very rapidly, and that rate of growth continued could swamp discussion. I wonder whether we might need to re-introduce a scenario that (IIRC) previously held whereby any new member had to have their first few posts approved before they became publicly visible.


Thoughts/insights?

Can't we taunt the bots? It's not like they have feelings.

 

[Armchair Cyclist]

Can't we taunt the bots? It's not like they have feelings.

In which case taunting them has no effect. If they can monitor response by volume, then I guess that replies are greater motivation. On that basis, please don't.

 

[scribers]

Can't we taunt the bots? It's not like they have feelings.

I would still find a way to lose badly

 

[Devil's Elbow]

I have asked Admins, and their approach seems to be that so long as there is not spam-links involved, they are unconcerned, and no rules are being broken.

Lol, of course that would be their approach. These are the times I wish we would just pack up and move the community elsewhere…

 

[Netserk]

The two main concerns I have are disruption and how it affects norms once people get to expect that new (more or less prolific) accounts are not humans. Long term, if bots are tolerated I don't think it will do anything good for the atmosphere here.

 

[Tim Cahill]

One of the main reasons I continue to post on here is because I know I'm talking to a real person, 15 years ago twitter was brilliant, now it's a dying platform, AI one of the main reasons for that.

That said I don't think we will have too much trouble, it's painfully easy to see who they are on a place like this and they really are few and far between by comparison.

 

[Potomac]

Look, if we can train these bots up on proper bantering it might be a good thing when they become our new overlords.

 

[tobydawq]

I have asked Admins, and their approach seems to be that so long as there is not spam-links involved, they are unconcerned, and no rules are being broken.

Wtf? This seems like a bizarre stance.

I have no interest in reading what a bot has to say so at the very least, if their posts are removed I don't have to waste time reading their crap. But that won't happen from now on??

Also, given that I was scolded yesterday for replying sarcastically to a bot, it seems rather weird that their posts will be allowed to stay but with the rule that we can't reply to them. That is not a sustainable situation and would give you much more work than if you just removed bot posts.

 

[Armchair Cyclist]

Wtf? This seems like a bizarre stance.

I have no interest in reading what a bot has to say so at the very least, if their posts are removed I don't have to waste time reading their crap. But that won't happen from now on??

Also, given that I was scolded yesterday for replying sarcastically to a bot, it seems rather weird that their posts will be allowed to stay but with the rule that we can't reply to them. That is not a sustainable situation and would give you much more work than if you just removed bot posts.

Which is why the most obvious ones, most off topic (like the one you engaged with) I am deleting, and banning if they are repetitive; and why I am asking people to report rather than engage,

 

[tobydawq]

Which is why the most obvious ones, most off topic (like the one you engaged with) I am deleting, and banning if they are repetitive; and why I am asking people to report rather than engage,

Okay, I thought you would begin to adhere to the admins' apparent wishes of letting them be. Fine then.

 

[Red Rick]

The two main concerns I have are disruption and how it affects norms once people get to expect that new (more or less prolific) accounts are not humans. Long term, if bots are tolerated I don't think it will do anything good for the atmosphere here.

I agree therefor I assume you're human

 

[Red Rick]

The obvious solution is to basically have the shadowban and need manual approving of new accounts until they meet a certain threshold.

The idea that as long as they don't post spam it's okay doesn't pass the sniff test for me cause it'll just get too much quickly.