Darth Lefty 

It's being rolled out in an advertising context so the answer to that should be obvious

ThermionicScott 

I've noticed that the hands in AI-generated art are getting better and better. It would be naïve to think this is all just being done for amusement.

HTupolev

No, but I might object to some equivocation in how you're leveraging the word "using." Generally when someone "uses" a traditional search engine to assist in constructing a forum post, for example, they're not assuming that the raw output from a preliminary search is suitable for direct use as that forum post.

Now, current LLM chatbots are maybe somewhat different due to how they correlate concepts, and how their primary output is natural-language text. I don't think there's anything inherently wrong with leveraging them to write forum posts. However, they're still in a preliminary state, and their ability to produce meaningful and appropriate answers is still fairly reliant on the user's subject understanding.

Another danger with the new chatbots is that the apparent coherence of their responses leads people to overly trust their rigor. For example, you're a strong advocate of using a chatbot to cite sources. But you've demonstrated it with a chatbot that isn't actually designed for this purpose, and which frequently hallucinates references to sources that don't exist.

livedarklions

Yes.

livedarklions

Even when the source is real, the AI's "interpretation" of the source's information can likewise be hallucinatory. A human needs to actually look at this source to see a) if it's quality; b) what it is saying and c) the context in which it is saying it.

I think the other thing that our friend here isn't acknowledging is that the AI can just overwhelm any rational response possibility by the sheer volume of junk it can produce at the push of a very few buttons. I believe that he and I have both illustrated that adequately.

IBJoel

Jeff, I'd just stop posting AI rsponses. We can't stop you, but if there's copyright changes that occur in the next year and we get takedown notices... I can tell you we're not going to sift through which responses of yours were generated or thought up and we'll just bulk-delete everything you've posted.

Darth Lefty 

@IBJoel we can't be the only IB forum,or indeed forums in general, talking about this. Do you foresee a policy or just let it ride for a while until the situation becomes clearer?

IBJoel

In fact, Bike Forums is. We haven't had this issue on other sites. So probably the latter option.

datlas 

Hi! What's this thread about?

tyrion

Hydraulic brakes vs. cable brakes.

shelbyfv

Just used ZeroGPT to confirm an AI post in Clydes. Thanks for the link!

livedarklions

There's been some reporting of publishers having to shut down online submissions as the volume of AI crap is too much to screen. I suspect you are going to need policies shortly, but I will be happy to be wrong. My preference would be to have a fairly strict stated ban, then rely on reporting by other users for enforcement. Obviously, it wouldn't be a ban on referring to it, it would be a ban on actual plagiarism and/or overlong quotation. I don't think that's substantially different from the copyright issues y'all have had to deal with for decades now.

Iride01 

One of the issues I've seen with chatGPT generated stuff is that it's happy to accept and include things that are considered dogma but still pose them as legitimate things that must be done.

So while the users might think they are gaining some education, it might be a dated education and not to current thinking.

unterhausen

Some of the spammers that come here love ai. They probably think it makes them less likely to be detected, which is not really true. It makes it just a little more satisfying to ban them.

I'm not going to catastrophize over people using AI. I swear some people on BF were in on the super-secret early beta program for it given the length and nature of their posts. We are generally discouraging it and have deleted some ai-generated posts by long-time users. Certainly not all of them yet.

ggpepper

I sometimes wondered about VintageSchwinn.

shelbyfv

That would be disappointing. I've always appreciated the effort that goes into those screeds. The derangement, not so much.