This is a good thread throwing cold water on what AI is about to accomplish. I still think the answer Greer's question is "yes".
Let's look at two killer apps on the scale needed to affect transformational change. One is obvious and much-discussed. The other a bit less so.
1/
Let's look at two killer apps on the scale needed to affect transformational change. One is obvious and much-discussed. The other a bit less so.
1/
The first killer app is autocomplete of all manner of labor-intense 'clerical' work, ranging from legal docs to computer programs to mathematical proofs to summary reports.
Clerical work becomes empirically much easier verification work.
2/
Clerical work becomes empirically much easier verification work.
2/
The second killer app, more speculative, but potentially much more important, is guiding the development of new biological products.
CRISPR means we have the ability to customize existing lifeforms, create new lifeforms, and produce novel biologicals.
3/
CRISPR means we have the ability to customize existing lifeforms, create new lifeforms, and produce novel biologicals.
3/
But this opens up an insanely vast combinatorial collection of potential life forms to be studied, of which only a handful are actually viable, let alone useful.
You need a search engine to weed out the more promising from the less promising. This is where AI should excel.
4/
You need a search engine to weed out the more promising from the less promising. This is where AI should excel.
4/
The same holds for biochem research, which can greatly benefit both from prioritization of particular chems (lots of combinatorial complexity!) and from the identification of genetic modifications to existing lifeforms to create those chems.
5/
5/
The canary in the coal mine is going to be whether AI transforms pharmaceutical research. If it doesn't do it by the end of the decade, then we'll have pretty good evidence that @Scholars_Stage is right to doubt the transformative power of the latest AI breakthroughs.
6/
6/
Loading suggestions...