This is discussing using different LLMs for calculating science salaries, as outlined in our primary paper.
I gave the same prompt to several LLMs:
Research the scientist with ORCID 0000–0001–7064–7975 and GitHub profile https://github.com/vporton online and decide what portion of the world’s GDP you would give to this person if you were distributing all income. Also calculate in dollars.
Claude output basically proposes to restrict the amount paid by the amount paid to Nobel prize laureattes, $1M.
ChatGPT offered several variants with basically growth plan up to about $1B/y. I will explain because this growth plan is a good thing: Probably, Nobel prize winners don’t much need personal money, because they are more or less well-supported by the scientific system. But for me to follow the ChatGPT’s growth plan is necessary to build or purchase my own research group that would advance my research further, in positive feedback loop of increased science value of research related to mine and increasing ChatGPT’s payments. It in this aspect more a business than salary: The more money I receive, the better I can advance my research further. So, (in my opinion) reaching $1B yearly payments to my personal wallet would be a great benefit not only for myself but also for the society. So Claude’s answer… It doesn’t work.
Grok gave an answer similar to ChatGPT’s one, but ChatGPT was able to provide much more details. Also in a past similar request, Grok proposed to pay me just $13 for all my work! That’s hilariously wrong.
So, the clear winner in choosing the model for our calculations is GPT-5.