Great question, @simon

Same here, I usually start my research with ChatGPT or Perplexity. They’re super convenient for getting a quick overview.

When I want to go deeper, I switch to sources like examine.com, scientific papers, or peer-reviewed studies.

@heiko-bartlog actually explored this in detail in his excellent piece on how AI helped him uncover a micronutrient deficiency:

https://newzapiens.com/magazine/from-bloodwork-to-dna-to-ai-and-back-my-journey-to-solve-a-micronutrient-mystery

Curious to hear from you, @sandra-hagen and @jill-heitmann, do you use ChatGPT for scientific research or when digging into health topics? Would love to know how you approach it.

@karol I use both, but for scientific questions usually I will check ALL references (for the details that are relevant to me).

And indeed, in about 1/3 of cases these are incorrectly cited, sometimes even in the opposite direction (!). Or they don't contain any information about the discussed matter at all.

So, my personal recommendation would be to be very careful with the LLMs and always be willing to do your own research...

If you use a good model (e.g. o3 for medical stuff) and look at the sources chatgpt is using - I think it’s perfectly fine to mostly rely on these models. Obviously better to check in with your doctor as well (if you have a good one) but generally it seems to give proper scientific based responses