Wissenschaftler, die einen zu 100 % automatisierten KI-Wissenschaftler entwickelt haben, behaupten, dass dieser bereits unabhängige Forschung betreibt, Entdeckungen macht und Artikel verfasst, die den Akzeptanzstandards wissenschaftlicher Zeitschriften entsprechen.
https://techxplore.com/news/2024-08-ai-scientist-scientific-autonomously.html
2 Kommentare
Submission Statement
Caveat – their claims have not been replicated or peer-reviewed yet. That said, I suspect they may well be true, or others will soon achieve the same thing.
AI making science discoveries has many pluses, but it adds to an existing crisis in human science. Academic appointments and funding are measured by the metric of published papers. There’s lots of evidence the system is gamed and corrupt. Many papers‘ claims are dubious, fail to replicate, and worse still – no one has the time or manpower to check them all. Maybe that can be a job for AI?
To check if the idea hasn’t been researched yet, it has to scrape all relevant research first. I don’t see that step.