Talk Morgan Forums Community The Malvern Arms AI trainer

Just discovered such on a Royal Enfield bike forum I belong. A new registered "member" asks on their first post a cure to a plausible issue on a particular model of vehicle but with only brief details. The posts flood in with solutions but no further from the OP. The AI company that created the new registration is then able farm the forum to benefit their product by offering a digested version of the common answers to their users.

I wonder what safeguards there are against the likelihood of such being attempted on TM in the future, if it hasn't already

There are various ‘systems’ in place to help protect the forum against a variety of nefarious registrations and activities. I won’t discuss them on the open forum because that helps those so inclined to circumvent them; however, if you’d like to know what they are, I’m happy to chat over PM



With that said, it also isn’t new for people to join, only to ask a question, or try to sell something and to never be heard from again. Little we can do to stop the first part but the later, to do with sales, is the reason I have a time restriction on sales posts and they all need to be approved.

Thanks Simon. Admit to being naive to think this is something new, but AI as a refined tool seems to be suddenly everywhere.

To be honest Richard, in my limited understanding of the way AI is being used by the major players, they don't need to register to harvest the information they require to improve the knowledge base of their programs. Just like the 'spiders bots' that constantly scan for new content, AI programs do the same but as far more clever with how they process what they find. Although sections of TM are closed unless the viewer is registered, there is a lot of information that is constantly being scanned because it is open to guests as well as users.

As I type this, there are 12 registered members online but 75 guests and 8 'bots' (the majority being Bing (Microsoft), 2 from Google and one from DotBot)

Saw a report that it had been judged that it is not possible to copyright the AI video or pictures generated by text prompts ... so you can have them generated for you by your AI of choice, but any monetisation would have to be pretty darned swift before the 'copy-jackals' who trawl the web for interesting content then immediately repost it under another name to skim the clicks and revenue sucked the value out of it. No comeback.... Easy-come easy-go in action I guess!

K

Here is a simple example image you can use to help you learn how to identify Bot driven images. Four classic failures on this AI generated image have been ringed in red to show where the errors are often made. See if you can work out how I spotted it.



Stolen from a Farcebook post by a more skilled person.

I found that image scary I'll admit - just too good and a bit like Katherine, I'll bet William wouldn't want them on him for a warm up in bed..

Ask AI a frivolous question and it will reply with a serious answer (plus a smiley).

What did my neighbour have for breakfast ?

I can't know that! But you could ask them directly or check for clues like food smells or dishes in their sink. 😄

A wonderfull example is the "Google Gemini" spot in the German TV.

in this spot a man ist situated und his car an says "Gemini how can I loosen the cat".

The answer ist: "You need a ratchet spanner or wrench that fits."