I still don't see the use case for mediocre research. The world doesn't progress on the basis of summaries of information which is mostly wrong (as all original research is) or obvious.
Nevertheless, I'm currently more sanguine about AI than I've ever been before. Some of its applications, particularly in the softer natural sciences like biotech and agriculture are going to lead to extraordinary breakthroughs.
On the other hand, I agree that probably won't come close to justifying the capital expenditure that has already been made and committed. As well as the potential surpassing or replication of current cutting edge models, currently state of the art hardware is going to rapidly become obsolete. There were many people sagely comparing NVidia to a company selling shovels in a gold rush who didn't bother to investigate just what proportion of spending goes on shovels at today's most productive gold mines. I'd rather have shares in Komatsu than Cyclone.
They’ll change quite a bit. but I don’t see any sign of the radical transformation being claimed. No real growth in measured productivity, no large scale displacement of workers being attributed to AI etc. As with computers and then the Internet, the gains will come as people learn to use these tools, but that will take a while/
Of the various bits of AI, I think LLMs and image generators have probably reached the point of diminishing returns to computing effort and have exhausted the data sets on which they operate.
I’ve followed your advice and for the history I am writing the research one is very good at giving the basics of what happened in a specific context (e.g Nasser, Tito, Nehru, and Zhou’s relationship) to enable a google scholar search.
I still don't see the use case for mediocre research. The world doesn't progress on the basis of summaries of information which is mostly wrong (as all original research is) or obvious.
Nevertheless, I'm currently more sanguine about AI than I've ever been before. Some of its applications, particularly in the softer natural sciences like biotech and agriculture are going to lead to extraordinary breakthroughs.
On the other hand, I agree that probably won't come close to justifying the capital expenditure that has already been made and committed. As well as the potential surpassing or replication of current cutting edge models, currently state of the art hardware is going to rapidly become obsolete. There were many people sagely comparing NVidia to a company selling shovels in a gold rush who didn't bother to investigate just what proportion of spending goes on shovels at today's most productive gold mines. I'd rather have shares in Komatsu than Cyclone.
I take it you're either skeptical that agents will "work" or you think they won't change much?
They’ll change quite a bit. but I don’t see any sign of the radical transformation being claimed. No real growth in measured productivity, no large scale displacement of workers being attributed to AI etc. As with computers and then the Internet, the gains will come as people learn to use these tools, but that will take a while/
Of the various bits of AI, I think LLMs and image generators have probably reached the point of diminishing returns to computing effort and have exhausted the data sets on which they operate.
I’ve followed your advice and for the history I am writing the research one is very good at giving the basics of what happened in a specific context (e.g Nasser, Tito, Nehru, and Zhou’s relationship) to enable a google scholar search.