Looking through different sections in the Seabreeze forums I am starting to wonder if a lot of the recent topics are posted by AI bots. Some of the topic starters are very specific questions yet the followup conversation is lacking knowledge of specific terminology. In addition the topics are almost always started by new users.
Thoughts?
I think they tested bots here for a loonngggg time so this is unsurprising. Some were pretty good and sounded like they knew stuff then you recognise bits from other posts here
Always ask yourself: what's in it for them? Spamming, scams, and advertising, yes. Asking questions about windsurfing, doubt it. Who would be driving that and for what purpose? Aka how would anyone monetise that?
I think it's more likely humans that post these things and just never come back to the forums after the initial post.
You would think that they wouldnt test this on us smart bunnies. Wingers maybe, kiters definitely, but not us.
Most AI are also easy to recognise in a forum like this as they write too perfect. No grammar mistakes, perfect punctuation, no typos, etc. Also, they often write a topic and titles in Title Case (capitalising the first letter of each word).
Eg.
How Can I Get Better At Gybing
That would be a sign that it might have been generated by AI.
Always ask yourself: what's in it for them? Spamming, scams, and advertising, yes. Asking questions about windsurfing, doubt it. Who would be driving that and for what purpose? Aka how would anyone monetise that?
Never underestimate a clever 13 year old with lots of spare time, no money, and a computer.
Most AI are also easy to recognise in a forum like this as they write too perfect. No grammar mistakes, perfect punctuation, no typos, etc. Also, they often write a topic and titles in Title Case (capitalising the first letter of each word).
Eg.
How Can I Get Better At Gybing
That would be a sign that it might have been generated by AI.
Don't be fooled by this sort of thing. Like everything, it can be changed to fool people to look more life-like. It would be trivial to add in spelling mistakes so that it looked more human.
I think the best way to detect these things is to look at what they are saying and seeing if it makes sense and if its from a newly joined user. Sometimes the wording is too generic and you can assume it is not a genuine question.
Do you have an example of such question in this forum? I still don't see why anyone would be wasting their time and resources on this. Running AI is not cheap. A teenager could run a bot on their computer but for anything serious they need to spend $$$ which they won't do unless theirs something to gain.
Would be happy days if teenagers are interested in windsurfing again.
I had AI generate some questions with intentional errors for your entertainment. Also notice the use of Title Case in the options.
Option 1: Slightly Rushed/Typo-Prone Human
"Hey their! I've been really trying to nail my windsurfing gybes lately, but I feel like I'm hitting a platue.
What's one tip that truly helped you make a breakthrough with your gybe, especially in getting a smoother, more consistant exit?"
Option 2: Casual/Less Formal Human
"Yo! I bin trying to nail my windsurfing gybes lately, but I feel like I'm hitting a wall.
Wut's one tip that like, really helped u make a breakthrough with your gybe, 'specially in gettin' a smoother, more consistant exit?"
Option 3: Minor Omissions/Grammar Hiccups Human
"Hey there! I've been really trying to nail my windsurfing gybes lately, but I feel like hitting a plateau.
What one tip that truly helped you make a breakthrough with your gybe, especially in getting a smoother, more consistent exit
I had AI generate some questions with intentional errors for your entertainment. Also notice the use of Title Case in the options.
...
Wut's one tip that like, really helped u make a breakthrough with your gybe, 'specially in gettin' a smoother, more consistant exit?"
A human that cuts down words and then uses punctuation to show this? Is there such a thing? Surely that would tip you off to it not being human.
I agree that it seems pointless to do these sorts of things, but some people get kicks out of it for some reason. Even Adolph here will use generic AI generated questions every now and then and you have to wonder 'why are they doing it?'. I could be wrong on that and he is just asking stupid questions.
On a car forum I was using recently there was a question posted that was identical to one someone had posted years before. Someone saw it and called them on it, but why would someone even do that? My best guess is that they are automating things and just trying out different scripts to see what works.
Do you have an example of such question in this forum?
The one I spotted has now been deleted. I'll see if I can find another example.
Running AI bots that engage in forums at scale is actually quite cheap in the scheme of things. There is a multitude of reasons. One very simple one could be to avoid a content harvester from being blocked.
Yeah that could make sense for forums that have member-only sections. Seabreeze has bot protection via Cloudflare so it should stop some automated abuse.
I think it can be useful to have accounts at hand for future use, eg to spam as soon as the account has reached a certain level that allows for posting images.
I don't think it's pointless. As I said, a few years ago it was very obvious here that new user would aggregate lots of things from threads and answer questions. They were clearly bots - so I assume testing/learning
More to the point is AI scary AF? I think so. The faked videos with no human input that look like a news story are not easy to spot. The person is rational - people are not. Look at the dunny paper fiasco during covid just because a news service reported it was getting low. an AI story about a disaster, war, politics etc could be really unhelpful .. Then it fact checks for us doesn't it.?
And FB is intentionally inserting "AI Agents" into groups (at least they are opt in). Purely for driving engagement.
In one of my local sailing groups, it started spamming with things like "How do you get over capsizing?"
I promptly blocked the thing.
It's inevitable in big corporate spaces like FB because we aren't the customer, we're the product. Seabreeze and local forums where they still exist have less of an incentive to do this so they will remain less affected for the near term.
/rant
The same FB that modified people's posts to see how people inferred mood, and if they could affect group mood (mass psychology / group culture )
that they got in trouble for
yeah, we can trust them
PTSF1111 said: No grammar mistakes, perfect punctuation, no typos, etc.
Personally, I find it impossible to write without all of these being correct, as each is what I've been taught over the last 60 years or so.
Even my phone texts to friends are correct.
Objection finished ![]()
Do the AI bots understand the difference between; "Helping your Uncle Jack off a horse" and, "Helping your uncle jack off a horse".
I had AI generate some questions with intentional errors for your entertainment. Also notice the use of Title Case in the options.
Option 1: Slightly Rushed/Typo-Prone Human
"Hey their! I've been really trying to nail my windsurfing gybes lately, but I feel like I'm hitting a platue.
What's one tip that truly helped you make a breakthrough with your gybe, especially in getting a smoother, more consistant exit?"
Option 2: Casual/Less Formal Human
"Yo! I bin trying to nail my windsurfing gybes lately, but I feel like I'm hitting a wall.
Wut's one tip that like, really helped u make a breakthrough with your gybe, 'specially in gettin' a smoother, more consistant exit?"
Option 3: Minor Omissions/Grammar Hiccups Human
"Hey there! I've been really trying to nail my windsurfing gybes lately, but I feel like hitting a plateau.
What one tip that truly helped you make a breakthrough with your gybe, especially in getting a smoother, more consistent exit
Did you ask it to ask for "one tip"?
Thats what makes it stand out for me. Humans won't ask for "one tip". They'll want whatever tips they can get. Like my mate after he separated for his missus. His FB profile changed from "married" to "whatever I can get".
I asked for a question, not specifically one. Good point and most people that start a thread would also add some context.
We can build a bot that rates new posts with a humanity score ![]()
More and more i'm feeling that nothing on screens can be trusted fully. Things online these days clearly can have heavy effects on society, ie 'The Great Zombie Toilet Paper Stampede of 2020' lest we forget. An insane heard mentality response of mass proportions.
The words 'covid tax' kept coming up everywhere all of a sudden that i was seeing. Chat bots could do that to affect the market i guess. I know it did, used car values for one example seemed to drastically change. But i wasn't buying a car then, i was busy collecting cheap windsurfing gear ![]()
I don't know enough about video to know what's capable there, but again, screens, pixles, me trust not. I think fake news articles mimicking certain peoples faces could be created.
Apparently there's a Jim's WIndsurfing youtube channel that is an AGI been running for awhile. It looks like a guy that claims it learnt to gybe in a few years. Yeah right, inhuman. It's 6am, dark, slightly windy, i gotta go to work programming the thing to do double hand drags on the foil NOW!
I think fake news articles mimicking certain peoples faces could be created.
Oh dear. I think you're a bit behind man. This is the stuff I'm talking about. Were this 'news' it could create hysteria. Computer people do it cos its 'cool' or some crap without thinking what they're doing to the world....
Do the AI bots understand the difference between; "Helping your Uncle Jack off a horse" and, "Helping your uncle jack off a horse".
Mate I don`t even know what`s the difference .
I just assume you`re pissing off the horse by asking it stupid questions like .... " what`s wrong , why the long face ? "
Do the AI bots understand the difference between; "Helping your Uncle Jack off a horse" and, "Helping your uncle jack off a horse".
Actually they do... they would define this as an ambiguous response and would ask for clarification. Exact response depends on how the system is weighted.