[jola]

We have had a few worrying incidents recently with AI generated posts. Sometimes it has been pure bots, and sometimes it has been real persons without any background in Computational Fluid Dynamics, that have used AI to engage in advanced scientific discussions with our senior forum users.

These bots and fraudulent posters often have fake identities, using false academic emails, posting reference to their own publications in fake journals that they control, or linking to their own unreviewed and often false scientific papers that they have uploaded to free repositories like arxiv.org or zenodo.org. Many have also created false academic profiles on platforms that allow users to enter credentials without verification, such as Google Scholar, ORCID, ResearchGate, Academia.edu and LinkedIn.

The latest incident was especially worrying, and made me realize that scientific development, professional scientific discussions and using journal references is going to require a new awareness in the future. What is real and what is fake will require you to really review credentials and only trust references that are from well-known publications or conferences with an established peer-review process.

I am thinking about adding these new rules to our forum:

AI Content & Fake Identity Rules
To protect the integrity of scientific discussions, the following are prohibited:

• Excessive self-promotion and repeated links to your own unreviewed papers, particularly those generated with AI and uploaded to open repositories.
• AI generated plots without providing full access to the real data used.
• Fake identities: False claims of academic or professional credentials.
• Fake journals or conferences: References to self‑created or unverified journals or conferences used to simulate legitimacy.
• Flooding discussions: Mass posting of unverifiable or misleading material will result in an immediate ban.

What do you think about this?

[LuckyTran]

Fake credentials, amateurs masquerading as experts, and bad-faith actors are the most problematic issues that we need to eliminate. Genuine inquiry is always welcome, which is the primary content on this site anyway


To make it clear that there is not an outright ban on the use of AI, there should be several concrete examples of where AI use is permitted.

• Making a post/thread asking for further clarification on a topic should always be 100% allowed even if the post is just dumping their chat log with the AI
• Use of AI-tools to analyze data or perform simulations in novel ways is of course allowed even if they result in lackluster findings: such as using ML to again predict pressure drop in a circular pipe for the billionth time
• Using AI to help you express your ideas and make graphics are welcome. However, AI-generated plots need to utilize real data and display them accurately. No hoaxes. If you can't provide the raw data in a excel sheet like a .csv or a dataframe that can be loaded in python, then it didn't happen.

[FMDenaro]

"Mala tempora currunt sed peiora parantur"

[JBeilke]

Prohibitions have never been effective. We could establish a separate subforum for ML and AI. Then the bots could discuss matters there, and anyone interested could read along. Perhaps something truly interesting would emerge that we would not have thought of ourselves.

[jola]

Originally Posted by JBeilke

Prohibitions have never been effective. We could establish a separate subforum for ML and AI. Then the bots could discuss matters there, and anyone interested could read along. Perhaps something truly interesting would emerge that we would not have thought of ourselves.

Without any blockage of spammers, 95% of all posts here would be spam, so some form of prohibition is necessary I'm afraid. My post is not mainly about new forum rules, but more to make people aware that this is happening, and that we have seen users using AI together with fake references in fake journals engaging in scientific semi-discussions, perhaps with the main purpose of creating another online reference...

I'd also like to get some form of discussions going how the forum seniors want us to handle AI posts. Using AI to help when writing a post is perfectly okay, using AI to generate fake science with fake references and fake credentials is not okay in my humble opinion.

The AI bots will of course ignore the forum rules, just as all spammers ignore our rules. I wish we could start an AI forum . But do you have any clue how to get the AI-bots and pure AI driven posts to stay in one dedicated forum?

[jola]

Originally Posted by LuckyTran

Fake credentials, amateurs masquerading as experts, and bad-faith actors are the most problematic issues that we need to eliminate. Genuine inquiry is always welcome, which is the primary content on this site anyway

To make it clear that there is not an outright ban on the use of AI, there should be several concrete examples of where AI use is permitted.

I agree with you. Perhaps my original suggestion is a bit too AI negative. AI is a fantastic tool, that if used in the right way can benefit scientific discussions. The problem is to avoid the problems that follow with the power AI brings to everyone. AI can very easily be used to create fake science based on fake references and invented data.

[sharonyue]

I do agree with Jonas. Indeed AI can answer lots of questions. But AI create fake science! and the CFD beginners are not aware of it at all.

But some people would not be happy. They are using AI to do research. You would be against with them if you say "AI bring fake science", because they used AI to write papers. In my Chinese media, I tried to make people aware AI's fake science. I run into objection.

Last week I found this paper "Global Regularity of the Three-Dimensional Incompressible Navier-Stokes Equations via Arithmetic Viscosity on the Leech Lattice". The author said "WE DID IT! The Millennium Prize 11/26/2025". I am not sure if its an AI bot yet. It is getting harder and harder to tell.

[jola]

Originally Posted by sharonyue

The author said "WE DID IT! The Millennium Prize 11/26/2025". I am not sure if its an AI bot yet. It is getting harder and harder to tell.

This is a problem we have just seen the start of. With AI it is easy for people without the proper scientific background and without any experimental validations to mass-produce papers that look very professional when you quickly browse through them.

Then you can upload these unreviewed papers to free paper archives like arxiv.org or zenodo.org, and create professional looking profiles on Google Scholar, ORCID, ResearchGate, Academia.edu and LinkedIn.

If you do this on an interesting problem, for example a Millenium Prize problem, you have created a very good click-bait. Then you can start talking about it on scientific sites like CFD Online to drive up the number of downloads of your unreviewed and possibly fake papers.

I am not sure how a site like CFD Online should handle this. Should we forbid people from posting references to their own AI generated papers that have not been reviewed by anyone?

I do not want to forbid innovation, but we do not want a forum filled with a lot of semi-discussions based only on references to unreviewed AI generated papers.

[jola]

I have updated my original post and adjusted my proposed new forum rules about AI Content & Fake Identities