The use of AI is growing in my field rapidly (meteorology). A lot of AI forecasts are doing well, even better than some of the traditional models that are based on mathematical equations. However, the AI models are trained on the data created by the traditional models so we wouldn’t have them without decades of research that went into those!
There’s a lot of hype in medicinal chemistry regarding AI – companies have been set up on the promise of much faster discovery of drugs. I think in reality AI has a role in mining “big data” which allows for the design of new potential medicines, however, the physical properties of a new compound and how the body sees it and deals with it (removes it) are a very large part of a successful new medicine which I don’t think is predictable yet.
The other problem is that much of the data for new medicines is in patents and is not uniform, so difficult to combine into a big data set.
Comments
Mark commented on :
There’s a lot of hype in medicinal chemistry regarding AI – companies have been set up on the promise of much faster discovery of drugs. I think in reality AI has a role in mining “big data” which allows for the design of new potential medicines, however, the physical properties of a new compound and how the body sees it and deals with it (removes it) are a very large part of a successful new medicine which I don’t think is predictable yet.
The other problem is that much of the data for new medicines is in patents and is not uniform, so difficult to combine into a big data set.