im skeptical but open to that. it’s just that these models are pushing pushed into literally everything, to the point they’re hard to avoid. I can’t think of another kind of specialized lab tool that has had that done. I do not own, nor have I ever owned, a sample centrifuge. I don’t have CRISPR tools. I have never, outside of academic settings, opened wolfram alpha on my home computer. even AUTOCAD and solidworks are specialist tools, and I haven’t touched any version of either in years.
because these models, while not good for anything anyone should ever actually want outside a lab setting, are also very very good for fascism. they do everything a fascist needs to, aside from the actual physical killing.
and I don’t think the level of development and deployment that these tools get, along with the wildly inflated price of the hardware to run them (or anything else) and death of web search, the damage to academic journals, etc, is a net benefit. even to specialized researchers who have uses for specialized versions of them as the statistical tool that they are. certainly not to the fields over the long term.
Why shouldn’t they have long term benefits for researchers?
Reminds me a bit of when CRISPR got big, people were worried to no end about potential dangers, designer babies, bioterrorism (“everybody can make a killer virus in their garage now”) etc. In reality, it has been a huge leap forward for molecular biology and has vastly helped research, cancer treatment, drug development and many other things. I think machine learning could have a similar impact. It’s already being used in development of new drugs, genomics, detection of tumours just to name a few
because murdering truth is not good for science. fascism is not good for science funding. researchers use search engines all the time. academia is struggling with a LLM fraud problem.
I mean yes, but it can be VERY useful in these narrow laboratory use cases
im skeptical but open to that. it’s just that these models are pushing pushed into literally everything, to the point they’re hard to avoid. I can’t think of another kind of specialized lab tool that has had that done. I do not own, nor have I ever owned, a sample centrifuge. I don’t have CRISPR tools. I have never, outside of academic settings, opened wolfram alpha on my home computer. even AUTOCAD and solidworks are specialist tools, and I haven’t touched any version of either in years.
because these models, while not good for anything anyone should ever actually want outside a lab setting, are also very very good for fascism. they do everything a fascist needs to, aside from the actual physical killing.
and I don’t think the level of development and deployment that these tools get, along with the wildly inflated price of the hardware to run them (or anything else) and death of web search, the damage to academic journals, etc, is a net benefit. even to specialized researchers who have uses for specialized versions of them as the statistical tool that they are. certainly not to the fields over the long term.
Why shouldn’t they have long term benefits for researchers?
Reminds me a bit of when CRISPR got big, people were worried to no end about potential dangers, designer babies, bioterrorism (“everybody can make a killer virus in their garage now”) etc. In reality, it has been a huge leap forward for molecular biology and has vastly helped research, cancer treatment, drug development and many other things. I think machine learning could have a similar impact. It’s already being used in development of new drugs, genomics, detection of tumours just to name a few
because murdering truth is not good for science. fascism is not good for science funding. researchers use search engines all the time. academia is struggling with a LLM fraud problem.