The future of the human brain — A review

Written by Sahit Jayakrishna, Senior Undergraduate, IIT Gandhinagar

The human brain is a fascinating organ and there is no area unlike the field of Computer Science which is more inspired by it. The entire field of Artificial Intelligence having been built around the concept that one day we will be able to teach machines to be able to teach themselves to think and make decisions like humans. However, as surprising as it seems, we are, quite frankly, pretty far off from that goal. That would probably come as a shock compared to what we keep hearing about AI. Although we have achieved a lot but that’s just the tip of the iceberg. There is still a long, long road ahead. We are, to a lot of extents, focusing on the area where machines come up with doing something human-like, for example, understanding knowledge.

Human beings have an astounding way of learning, which includes methods such as reinforcement learning or supervised forms of learning which also happen to be types of Machine Learning environments. But it’s not about the way of learning it, rather the way we store that information. In Computer Science terminology, it has to do with databases. We know how data is stored in databases (SQL and NoSQL), however, the human brains not only stores data but also knows how to easily inference from it.

An example always makes it easier to understand: Suppose we have information about a family and we are storing this in a database. Thus we will have details of the father, mother, children, uncles, aunts, grandparents etc. We can define a table with different fields catering to different sections. It is also possible to design a table where the computer can automatically compute the relations among different famil members the moment we feed in the information. This comes at the cost of a complex algorithm and more processing. As it also turns out this solution is not easily scalable to a larger amount of information. But think for a moment, our brain processes this information so effortlessly and without much hassle. This is what we are trying to achieve with today’s cutting-edge technology: to be able to mimic our brain with simpler processing and computes.

If being able to carry out inferencing from data is such a hard task, how do computers perform them with high precision? Simple — because one, for each task there is a whole new set of algorithms designed and two, computers are trained to find patterns, however not inference. One could argue that inferencing is in a way finding some patterns in the data which is true, but inferencing may not always be a subproblem of our main task we are trying to perform.

Take a standard example — Classification of fruits. If you were to ask an AI machine to do it, it would probably take a picture of it and try the best image classification algorithms, but as a human, you would probably take into account not how it looks, but also how it feels and smells maybe, something you know would work in this case, “an inference you made”.

The above example seems to also provide a solution for AI to be able to infer, which is like giving additional information. This is still a limitation that we are trying to solve. We would probably need to hook the machine up to the mind itself to get better at the task of inferencing, something which Elon Musk is already working on (Neuralink), but overall it seems we have a long way until we can make AI as capable as a human brain. With amazing and fast-paced advances in methodologies and tech, we might be able to crack the shell much quicker than what the trend and algorithms predict. Either way, for the time being, the human brain is still to stay.

Reference for the image:https://image.freepik.com/free-vector/artificial-intelligence-illustration-human-brain-suit-digital-mind_33099-556.jpg