Comment by uuuuuuuuuuuu

I feel like DeepMind has a disproportionately large scientific impact relative to its resource pool. How would one (or a group) go about replicating its success?