TuxBot and AI

DavePM

Member
Joined
May 2, 2023
Messages
95
Reaction score
37
Credits
771
Well I must admit I am blown away with how well the TuxBot works on this forum, I just never expected it to be so on point with its replies.

With this technology I suppose there are questions that we could ask and what it might do to our existing technologies we now use. Just take this forum would we really need a forum for anything if we just able to type in any question we might have into our desktop AI and get an instant answer that seems to be 100% correct most of the time.

Then it seems that even the scientific world is becoming concerned about what this technology is really capable of doing.

 


KGIII

Super Moderator
Staff member
Gold Supporter
Joined
Jul 23, 2020
Messages
10,143
Reaction score
8,715
Credits
83,882
I've paid some attention to the various threads and I'm really impressed with the bot's NLP capability. It's impressive how accurate it can be, uncanny really. I've seen the bot figure out what the OP was asking when I wasn't even sure what OP was asking. You can throw a lot of erroneous text at the bot and it still manages to figure out what you really meant to say.
 

wendy-lebaron

Active Member
Joined
May 3, 2023
Messages
340
Reaction score
102
Credits
4,553
Just take this forum would we really need a forum for anything if we just able to type in any question we might have into our desktop AI and get an instant answer that seems to be 100% correct most of the time.
Time to worry only if the bot were 100% correct most of the time. Because I had to correct TuxBot constantly in a thread I started in TuxBot's forum. Maybe it has profound knowledge about JavaScript or Python programming but never about an ancient seldom-used programming language. (shrugs)

It's fun reading how it answers sometimes. But most of its advice cannot be taken seriously. Must seek second opinions from humans that know, although that is hard. Getting opinions out of other people is the main reason why a forum exists such as this one. But a bot that must have every last word which resides on the forum is increasingly becoming attractive to both sides: people desperately looking for answers but want to avoid the "proud geeks", and the people that have to work on "customer support" whether or not they are paid for it.
 
OP
DavePM

DavePM

Member
Joined
May 2, 2023
Messages
95
Reaction score
37
Credits
771
Time to worry only if the bot were 100% correct most of the time. Because I had to correct TuxBot constantly in a thread I started in TuxBot's forum. Maybe it has profound knowledge about JavaScript or Python programming but never about an ancient seldom-used programming language. (shrugs)

It's fun reading how it answers sometimes. But most of its advice cannot be taken seriously. Must seek second opinions from humans that know, although that is hard. Getting opinions out of other people is the main reason why a forum exists such as this one. But a bot that must have every last word which resides on the forum is increasingly becoming attractive to both sides: people desperately looking for answers but want to avoid the "proud geeks", and the people that have to work on "customer support" whether or not they are paid for it.
I have a motorcycle forum and I went onto the demo site and actually posted motorcycle questions on a brand that my members had asked about in the forum itself. The answers that came back were amazing and at times even more accurate than what the live members had asked about.

I would love to add it to my forum but to date I just cannot seem to get any information on the pricing structure from OpenAI itself. So it seem rather up in the air to firstly purchase the forum software interface that runs here and then only after that finding out that the monthly price for the engine itself it far more than what I would like to pay.
 

sphen

Well-Known Member
Joined
Dec 12, 2022
Messages
868
Reaction score
750
Credits
10,394
I have not been impressed with TuxBot. Sorry. The following is my own personal opinion, based entirely on my observations of TuxBot on Linux.org. This is my own writing by my own hand:

Where TuxBot excels is its natural language processing. It communicates with people very well. I am impressed with its "casual friendly" natural style of writing, which puts the real person on the other end at ease. That is a tour-de-force of engineering.

On Linux.org, TuxBot suffers from its inability to evaluate the quality of its responses. It does not filter itself well to limit how much it posts bad advice, results, suggestions, and recommendations. I think that TuxBot quality will achieve the greatest improvement if the developers focus heavily on eliminating bad responses, rather than improving the good ones. This is a personal opinion.

I am concerned that TuxBot is polluting the internet with bad information. It can generate a lot more bad information a lot faster than real people with experience and knowledge can create their own or fix what TuxBot and its ilk does. TuxBot accelerates the entropy increase in the world.

Finally, the TuxBot implementation on Linux.org needs improvement. I will not post corrections to erroneous or inaccurate posts in TuxBot threads because I know I will get an annoying reply. My attempts to use TuxBot as designed have not been worth the effort. I am pleased that TuxBot forum threads no longer appear in "New Posts". That said, the current way that TuxBot operates is good enough if I avoid the TuxBot forum thread. I appreciate the attempts to make it better.

PLEASE READ:
It is easy to pick at nits, but I want to acknowledge all the hard work that @Rob and the team have invested to make TuxBot happen. It continues to be an interesting experiment and obviously adds value to @DavePM and others. I hope this feedback helps.
 

Brickwizard

Well-Known Member
Joined
Apr 28, 2021
Messages
4,697
Reaction score
3,167
Credits
34,919
It continues to be an interesting experiment and obviously adds value to @DavePM and others. I hope this feedback helps.
I think he does a reasonable job. But must remember he is a screen scraper.
And so presents the textbook answer, unlike most of us miserable old sods, Tux doesn't have the capability of lateral thinking, which is the way many of the problems we are faced with are solved.
 

KGIII

Super Moderator
Staff member
Gold Supporter
Joined
Jul 23, 2020
Messages
10,143
Reaction score
8,715
Credits
83,882
Instant access to information: AI language models can provide quick and accurate answers to a wide range of questions, eliminating the need to browse through forums or search for information manually. This convenience can streamline the process of acquiring knowledge and finding solutions.

Hmm... Are you a bot or maybe had AI write that response? At first blush, it appears to be one of those.
 

sphen

Well-Known Member
Joined
Dec 12, 2022
Messages
868
Reaction score
750
Credits
10,394
I think he does a reasonable job. But must remember he is a screen scraper.
And so presents the textbook answer, unlike most of us miserable old sods, Tux doesn't have the capability of lateral thinking, which is the way many of the problems we are faced with are solved.
TuxBot does not yet do a good job separating the old, obsolete, out of date "textbooks" from the newer ones with the current and correct responses.

We eliminate those older textbooks to avoid misleading our young people. We do not want them to learn arithmetic the old bad way any longer. Schools sell those nasty old textbooks to the recycle center by the pound.

Hmm... Are you a bot or maybe had AI write that response? At first blush, it appears to be one of those.
My thought exactly. New member. Posts in a fading thread after more than a week with no action. Post seems generic and is unrelated to Linux (but neither is the thread). My first thought was either spammer seeding or bot doing the same.

... And a member awarded a "Like" to it, too.
 

KGIII

Super Moderator
Staff member
Gold Supporter
Joined
Jul 23, 2020
Messages
10,143
Reaction score
8,715
Credits
83,882
My thought exactly.

Meh... Let us deal with it and hit the report button if it looks like it is. However, note that I didn't do anything other than ask them a question. Their comment doesn't actually violate any particular rules.

We do have a few generic rules that such comments could fall under, but those are things like "We may remove or modify any Content submitted at any time, with or without cause, with or without notice." There's also, "We may terminate your access to all or any part of the Service at any time, with or without cause, with or without notice."

I don't exercise those rules. I only act when it violates our other list of rules, which can mostly be summed up as not being a jerk and being welcoming. (Plus no politics, religion, discussion of sexuality/sexual in nature, etc. Which, again, could be simplified as to mean 'acting like you're in polite company'.)

That's enough of that topic for now. You can now resume bashing the bot or praising the bot in accordance with your thoughts on the matter. I do hope someone's reading those threads now that they're not in the list. I keep forgetting to check for moderation purposes.
 

Rob

Administrator
Staff member
Joined
Oct 27, 2011
Messages
1,181
Reaction score
2,172
Credits
3,302
Meh... Let us deal with it and hit the report button if it looks like it is. However, note that I didn't do anything other than ask them a question. Their comment doesn't actually violate any particular rules.

We do have a few generic rules that such comments could fall under, but those are things like "We may remove or modify any Content submitted at any time, with or without cause, with or without notice." There's also, "We may terminate your access to all or any part of the Service at any time, with or without cause, with or without notice."

I don't exercise those rules. I only act when it violates our other list of rules, which can mostly be summed up as not being a jerk and being welcoming. (Plus no politics, religion, discussion of sexuality/sexual in nature, etc. Which, again, could be simplified as to mean 'acting like you're in polite company'.)

That's enough of that topic for now. You can now resume bashing the bot or praising the bot in accordance with your thoughts on the matter. I do hope someone's reading those threads now that they're not in the list. I keep forgetting to check for moderation purposes.
Yeah, i deleted it :)
 

KGIII

Super Moderator
Staff member
Gold Supporter
Joined
Jul 23, 2020
Messages
10,143
Reaction score
8,715
Credits
83,882
Yeah, i deleted it :)

I agree 100% and considered it, but I err on the side of caution.

We have some pretty good catch-all rules that cover it. It certainly looked AI written and like a bot.

I posted something the other day, during Le Mans, about how my lowly site had like 4500 attacks (by just one metric) in the previous 72 hours. I only see the tip of the iceberg here. At some point, that AI bot is going to be able to sound like a real human.
 

sphen

Well-Known Member
Joined
Dec 12, 2022
Messages
868
Reaction score
750
Credits
10,394
Meh... Let us deal with it and hit the report button if it looks like it is.
I typed a report on it ... twice ... but clicked the Cancel button before filing the report. After that, another member awarded it a "Like" and the thread took this course instead.

I have noticed an evolution in the patterns of the spam posts here since late last year. The content style looks more "AI generated" and my searches for the sources of spam text are coming up empty the majority of the times I try them. The spammers are not yet tailoring their responses to match the victim forums very well. Those "generic" responses still differ from real people if you take a moment to read them.
 

sphen

Well-Known Member
Joined
Dec 12, 2022
Messages
868
Reaction score
750
Credits
10,394
What is their point in doing this? What do they gain?
Some forums enforce rules about the minimum number of posts required from new users before they can perform routine actions in the forum without moderator review first. They may also enforce restrictions based on how long the account has been on the forum.

Spammers are forced to create posts in the forums to establish themselves first before they start spamming the forums or PM'ing members. They come, leave a few posts and allow them to age. When the spam starts, it appears to come from someone who has been a member for an "reasonable period", longer than a few days or weeks.

I call those initial posts "foothold posts" (my term). The "foothold posts" must survive the scrutiny of the moderators and members, so they must look sufficiently real that they won't be deleted or cause the new account to be deleted.

Much of the work is done by programmed bots and not real people. It is currently a battle of escalation and attrition. Adding AI to the mix leaves better foothold posts that are more likely to survive.
 

Staff online


Latest posts

Top