> For now we see through a glass, darkly; but then face to face: now I know in part; but then shall I know even as also I am known.

Agreed that de-anonymizing will become trivial. This will be a problem not only for bad actors in research, journalism, creative writing, etc., but for internet commenters who believed they'd done due diligence to remain anonymous, and even for research participants who'd expected anonymity when signing their consent forms. We're rushing headlong into even stranger days!

I believe the complete opposite - at least for internet commenters, one could easily use an LLM to generate one's responses.

I would prioritize looking for financial and corporate fraud. It would have a much bigger impact on society than looking for any problems with academic studies. If we can take down those people and bar them from ever having anything to do with finance in the future, I think that would have an important impact on the ethics and behavior of the next generation.

It seems to me that taking down fraudulent academic researchers and barring them from ever having anything to do with research in the future would also have a significant impact on the ethics and behavior of the next generation. If technology is lowering the barriers to fraud detection, why should it be applied to one sector over another?

That's a fair point. However, part of the politics of the situation is that the right end of the spectrum appears bound and determined to discredit any academic or fact-based professions. They also seemed dead set on protecting oligarch and oligarch-lite players from criticism or critical examination.

Academic fraud seems to be self-correcting as we have seen by the numerous reports of fraud and withdrawn papers; as a rule, the players express shame and remorse. Nonacademic fraud doesn't exhibit these characteristics and sometimes seems to be proud of the fraud.

These researchers work for corporations too.

and corporate researchers do commit fraud as well. Look at the number of drug trials that hide results that would interfere with the approval of a drug.

how would LLMs help with uncovering "bad studies"? I don't think LLMs are yet sophisticated enough to figure out potential plagiarism or Adobe Photoshop copy-and-paste (as mentioned in the article above).

Well you can train AIs to uncover image fraud and detect manipulation that is well known (imagetwin does this).

Presumably the same can be done for tabular data and genomic or "omics" data. At the very least statistical techniques can be used. I imagine high throughput imaging modalities will be the main target.

That being said the only way the -omics data will have utility is via AI models which are trained on some task... which present their own problems..

The OP paper is easily handled by GOFAI from the 1980s; it's just detecting similar images.

You don't need a language model. The effort is in collecting and chopping the data to find snippets to compare. NNs can help with that.

You could probably just train it by taking image segments and duplicating them artificially.

Newer fraud will probably use generative diffusion AIs to make "realistic western plots" on demand .. there's probably a paper in that!