6 ms·

Google Parent Launches Company Dedicated to Drug Discovery

5 points by infodocket 5y ago

dekhn 5y ago
Welp, my endgame has been reached! Many years ago, I joined Google with a mission: solve drug discovery to the point where better molecules got approved by the FDA and made a real difference without costing tons of money. I was convinced that Google's technology (ML and data processing) could be applied to many biological problems and that it would be 'obvious' what the improvements over the existing state of the art were. Eventually I managed to work on these problems full time with enormous resources but Google proper never saw the value in it- which makes sense, because Google is an ads company that happens to run a popular website, with the goal of funding long term singularity research. And I burned out over time because dealing with the Google leadership is incredibly painful. Verily was a big waste, it's a joke, they have not and will not do anything useful, while continuing to provide funding for projects like Debug which have no future. Calico is... a retirement project for Art Levinson and a bunch of other smart people. Maybe they're do something amazing, maybe not. my money is on "not". However, DeepMind is completely different. They are a company founded with a long time of solving hard general intelligence problems and using those solutions to change the world dramatically. I knew from the very first time I saw Demis present his work on Atari games at Google (just after DM had been aquired) and it was obvious what the long game was: biology. Look at Demis's background and the way they run their research labs. It doesn't hurt that DM convincingly solved one of the hard problems in biology: cystal structure prediction. Now, don't let anybody's press releases convince you they solved protein folding (different problem) and the work doesn't immediately help drug discovery (no matter how much breathless articles will claim that). Instead, I think DM will continue with their promising work applying transformers and other methods to biological data. It's been one of the most fruitful areas in science I've seen so far, easily solving problems that plagued industry and academia for decades. Hopefully, they will do so in a way that benefits humanity, while also recompensating them for their efforts.

5 points by infodocket 5y ago

Welp, my endgame has been reached! Many years ago, I joined Google with a mission: solve drug discovery to the point where better molecules got approved by the FDA and made a real difference without costing tons of money. I was convinced that Google's technology (ML and data processing) could be applied to many biological problems and that it would be 'obvious' what the improvements over the existing state of the art were. Eventually I managed to work on these problems full time with enormous resources but Google proper never saw the value in it- which makes sense, because Google is an ads company that happens to run a popular website, with the goal of funding long term singularity research. And I burned out over time because dealing with the Google leadership is incredibly painful. Verily was a big waste, it's a joke, they have not and will not do anything useful, while continuing to provide funding for projects like Debug which have no future. Calico is... a retirement project for Art Levinson and a bunch of other smart people. Maybe they're do something amazing, maybe not. my money is on "not". However, DeepMind is completely different. They are a company founded with a long time of solving hard general intelligence problems and using those solutions to change the world dramatically. I knew from the very first time I saw Demis present his work on Atari games at Google (just after DM had been aquired) and it was obvious what the long game was: biology. Look at Demis's background and the way they run their research labs. It doesn't hurt that DM convincingly solved one of the hard problems in biology: cystal structure prediction. Now, don't let anybody's press releases convince you they solved protein folding (different problem) and the work doesn't immediately help drug discovery (no matter how much breathless articles will claim that). Instead, I think DM will continue with their promising work applying transformers and other methods to biological data. It's been one of the most fruitful areas in science I've seen so far, easily solving problems that plagued industry and academia for decades. Hopefully, they will do so in a way that benefits humanity, while also recompensating them for their efforts.