The energy inefficiency of AI is a telltale sign that they're on the wrong track. The human brain does what an LLM does with three orders of magnitude less training while consuming fewer than 10 watts of energy.
Biological intelligence and AI are currently qualitatively different, and if AI engineers were smart (which they're not; their greed has overcome their reason), they would attempt to discover biological intelligence's secret and improve on it.
Good science is an inherently socialist enterprise; bringing capitalism into the mix is always sure to spoil it.
And this phenomenon takes place from the largest scale (e.g., which kinds of research funding agencies and corporations are willing to fund) to the smallest, down to the interactions between individual scientists.
My collaborator and I have been victims of this lately. We developed some software. A couple of colleagues in my department recognized that it might have monetary value, so they plagiarized it and submitted the plagiarized code repository to the NSF as their own work.
My collaborator and I have been forced to deal with this situation for over a year and a half, during which time we have gotten almost no science done.
As a side-effect of this criminality, my collaborator has been unable to publish her dissertation research for over a year. The culprits of the plagiarism and federal grant fraud are effectively holding the paper hostage.
