NEUROSCIENCE & ARTIFICIAL INTELLIGENCE
Hype is the killer of reason and goes antithetical to a logical successful creation. Hype does not believe in a long gestation period. It has to happen right then and there and its successful completion should be announced at the earliest. In the fast moving world of artificial intelligence, when Demis Hassabis speaks the world listens with caution and concern. He has proved himself to be the raining deity of the artificial intelligence world.
Demis Hassabis has been the poineer of artificial intelligence, since the creation of DeepMind in 2010 and later its merger with Google in 2014. He still maintains functional research independence with the compute and other logistic support of IT behemoth like Google. Compute specialised AI, logic based dependent on huge data bases and open ML/AI tools is not a straight clear cut future of AI. Contemporary programs are extremely narrow and can be tricked. The question is how can AI be taken to the next level?
Demis Hassabis, neuroscience doctorate himself feels reconnecting to the discipline will create a “virtuous cycle.” He is convinced of the fact that artificial intelligence cannot have an independent growth story outside of neuroscience. “Distilling intelligence into an algorithmic construct (could) yield into some of the deepest and most enduring mysteries of the mind.” This they define as a Win-Win situation and the only acceptable trajectory that Artificial Intelligence can take.
Demis says that human brain is only existing proof we have for this sort of general Intelligence, which the world is trying to construct. Neuroscience has to be used as a source of information for algorithmic and architectural ideas. This is the reason he studied neuroscience for his PhD. Intelligence itself has to be understood and that can happen only through human brain. But the researcher needs to know what he should look for in the brain itself. Embodied cognition is an area to be built upon, where system is able to build its own knowledge from the first principle and then create abstract knowledge from there. The final frontier happens to be imagination.
WITH SO MUCH OF COMMONALITY, THE RESEARCH PATH OF AI CANNOT BE UNILINEAR.