Key points from article :
The few drugs that succeed take an average of 10 years to reach the market and cost $2-$12 billion.
Generative adversarial networks can start with minimal instructions and produce novel outcomes.
Alex Zhavoronkov started applying GANs to pharmacology in 2014.
AI constructs molecules from scratch based on verbal description drug attributes.
The result is an explosion in potential drug targets and a much more efficient testing process.
In 2018, Insilico Medicine generated novel molecules in 46 days - including synthesis.
The first drug to result from this work, a treatment for hair loss, is slated to start Phase I trials by the end of 2020.
AI could also predict the outcomes of clinical trials in advance - saving more money.
Other scientists are using AI to identify new drug targets (i.e. binding sites).
A small protein with a hundred amino acids can produce a googol-cubed worth of potential shapes.
AlphaFold predicting correct folding of over half the proteins in a 2018 competition.