The strategy for Google is quite simple, if there is any field they find worth investigating, they will see if someone is already up for it and then, they acquire that company to bring the technology under its aegis. Such is the case with DeepMind, an AI-based company from London who opened up new dimensions in AI based research as well as applications. Google’s acquisition in 2014 has proved to be handy as since the acquisition, its AI team has defeated humans in Go, the most complex game in the world and it looks to go beyond such achievements and maximize its abilities by making most of human intellect. AI is understood as a bunch of mathematical formulations. However, it should not emulate human brain, but actually work and think like one.
AI cannot be thought of without applications, no matter how you work with it. For now, AI has relied more or less on deep learning, which in turn depends on a series of complex algorithms that produces structural similarities with human brain operation. Deep learning works in area like object identification or speech recognition, things that a child’s brain can process. However, AI is supposed to do much more than simple general tasks where complex cognitive functions are performed such as memory, inquisitiveness and memory. Deep learning is an improvement on machine learning itself, and hence, suffers from similar limitations. Mathematics, to put it simply, will not be enough to deduce the limit of AI.
Versatility is the key
Algorithms are efficient, they need to become versatile. Algorithms that can modify or rectify itself or take their course according to sudden changes are yet to be designed and this is precisely the aim. Hence, a peek into neuroscience and its many implications becomes necessary because understanding the complete functional aspects of brain and its logic is necessary for developing the sharpest possible AI. The cutting-edge technology, hence, is no longer going to be about some unintelligible mathematical expressions but about the intricacies of neuroscience. There has to be some kind of symbiosis between the two vastly different fields, precisely the kind of collaboration that can open new horizons in the world of science and technology.
The relation between the two
Researchers have found intricate relations between deep learning and reinforcement learning, where trial and error plays a great role. However, it is not simply such repeated attempts but also intuition that plays a huge role in improving the cognitive assessment on things. Hence, what needs to be done now is to start studying the cognitive development of a child. If it is possible to understand at which point the child starts using its more human qualities like intuition and then, track the development minutely, it is largely possible to reach a point in neuroscience when the secrets of human brain are unlocked and turned into algorithms to make AI smarter than ever. However, the obvious difficulty in this regard is to find common expertise in both fields. But surely, that day is not far away.