• Guest - w'd love to know what you think about the forum! Take the 2025 Survey »

Notice for All Forum Users - B10 Rule Change - Artificial Intelligence "Chatbots"

Status
Not open for further replies.

KennyA

Moderator
Staff Member
Moderator
Messages
4,111
Location
Northern Ireland, living in Yorkshire
Type of diabetes
Treatment type
Diet only
Dear all,

In recent weeks we have seen a number of posts which appear to have been generated by machine intelligence - "chatbots". These posts are created by an Artificial Intelligence (AI) program through a synthesis of material generally available on the internet. The content is therefore neither direct personal experience (although posters may claim that) nor are they recognised and verifiable sources (although again, they may make that claim). Members posting the material are certainly not describing a genuine experience.

It is also possible that members acting in good faith might repost something first seen elsewhere without realising that the material is from a chatbot. Please be careful when reposting and bear in mind that these machine-generated posts are designed and intended to fool the reader into thinking that it is coming from a real person.

We suspect that people may be using the forums to "train" chatbots to "advise" people with the intention of selling the "trained" chatbot as a commercial item. What we have seen already indicates that the chatbots make mistakes, including not differentiating between Type 1 and Type 2 diabetes.

We think therefore that this sort of post is potentially harmful, and will continue to delete them as we detect them. Forum members deliberately posting chatbot material with the intent to deceive will be banned. We have amended forum rules - https://www.diabetes.co.uk/forum/threads/community-ethos-forum-rules.50278/
- (specifically B10) to make this clear.


The Moderator & Admin Teams
 
I'd like to add that those AI generated texts can be quite hard to recognise.
So if you see a post which gives you this feeling of something not being quite right. please hit the report button, even if you're not sure!

Using the report button opens a pop-up window where you can explain your reason for reporting, so you can simply tell us that you're not sure but wanted to alert the mod team that something may be off with a member, and could we have a look at it.
 
I’m assuming this is confined to newish members without too many posts? Kind of scary we need to be wary as these people often are the ones seeking the most help with limited knowledge as yet making it doubly difficult to pick up on something “wrong“
 
Hi,

Please by all means respond to new members & say hello. Chat bots tend to dish out generic “advice” with no detail or personal experience.

Don’t be wary of a bewildered newbie. Just be more mindful of a post in the topic responding that may not feel right or even appear the member seems to have “cut & paste” from somewhere..?

Report it & we can do the rest.. The team have been vigilant for quite some time. We won’t mind if a member has reported something suspect & it turns out to be an error. We look into every report seriously..

You can (for instance.) report the reason by just typing, “bot?” & the team will understand the concern.
 
One of the issues for us is that describing how we've spotted them helps their controllers learn how better to avoid being spotted. We'd rather not help them. However the chatbots seen up to now could not be mistaken for the archetypal confused newbie.

Chatbots have advanced quite a bit in the last year, and if the development continues at the same rate problems might get worse, although some countries are starting to take action - Italy has 'banned' ChatbotGPT, one of the best-known ones, whatever that means in practice. We're fairly sure that what we've seen so far is people cut and paste posting chatbot-generated content, not chatbots posting directly and autonomously on the forum.
 
First Officer said:
Does this mean the Turing test has been passed?
Click to expand...
It seems so

During a live demonstration at Reading University, the computer convinced multiple human judges that it was in fact a 13-year-old Ukrainian boy, not a machine. This stunt made the algorithm the first ever computer to pass the famous Turing test — successfully convincing over a third of the jury of its humanity. Today, Eugene Goostman stands as one of five supercomputers to have beaten the famous test.
Copied from here
https://www.wildfirepr.com/blog/can...ade the algorithm,have beaten the famous test.
 
Just an update, ChatGPT is available again in Italy: I played with it yesterday (asked it to come with a list of typical names from a particular background).
 
Although I don’t really even understand any of this, it sounds horrific, what’s the purpose? Presumably to make money, but how?
 
So, I thought it'd be fun to ask your question to ChatGPT Its answer was quite long, so here's just part...

Q: Although I don’t really even understand AI, it sounds horrific, what’s the purpose? Presumably to make money, but how?
A: Regarding the monetary aspect, AI can generate revenue through various means:

a. Product Development: Companies can develop AI-powered products and services that meet market demands. For example, virtual assistants, chatbots, recommendation systems, and autonomous vehicles are all AI-driven products that have commercial value.
b. Data Monetization: AI relies on data, and companies can monetize the data they collect by leveraging AI algorithms to extract valuable insights. This can be done through targeted advertising, selling data to third parties, or providing data-driven services.
c. Process Optimization: By implementing AI-driven automation and efficiency improvements, businesses can reduce costs, optimize workflows, and increase their profit margins.
d. Consultancy and Services: AI expertise is in high demand, and companies can provide consultancy and services related to AI implementation, development, and maintenance. This can range from advising businesses on AI strategies to building customized AI solutions for specific needs.

One thing I know some have done is use a chatbot to create content for a new website quickly. You can then use another AI, such as Midjourney, to create images to go along with that content. So that becomes a super easy way to get a marketing campaign up and running, or even a political campaign. The pictures you might have seen recently with DJT as a fighter pilot or a Native chieftain (wearing a Medicine Woman necklace strangely enough) were AI generated.
 
Last edited:
Mods are bumping this thread as we’ve seen another increase in recent weeks of AI generated posts, we’d like to remind members that any form of AI isn’t allowed on the forums please see rule B10 of the forum rules & ethos


B10. A zero tolerance policy to harmful or wholly inappropriate posts will be applied. Recently diabetes-related material generated by "chatbots" has appeared on the internet. This is a machine-created synthesis and is therefore neither personal experience nor is it a recognised and valid source as required by forum rules. It is misleading and potentially harmful. Members who deliberately post chatbot material will as a minimum have the post deleted and other sanctions may be applied.

Thanks for your understanding and cooperation
The mod & admin team
 
Status
Not open for further replies.
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn More.…