Why are we obsessed about comparing machines with the human brain?
Why the obsession of comparing machine intelligence with what the human brain can do? especially given that computers today can do a whole lot of things in milliseconds that the human brain can’t do even in years!
How do computers work today?
For the most part, computers today use logic-based software (let's call it ‘conventional’ software). The way the conventional software works is that you program a discrete set of rules in the form of if-else statements or loops at the most basic level and the computer applies those rules on the inputs to give the desired output. Computers have gotten a lot faster over the years so they can now perform those discrete set of rules much faster and efficiently. Hence we can now create an extremely complex and large set of rules and get the results in milliseconds, seconds or minutes depending on the complexity and the number of those rules.
What are the limitations of conventional software?
As it turns out, even though the logic-based software approach has worked really well for a very wide variety of problems, it still sucks badly when it comes to problems that involve an extremely large number of rules. For example, let's pick the problem of understanding natural language. To make sense of a sentence in any language you have to not just consider all the formal rules of the grammar but also the colloquial usage, the poetic usage, sarcasm and many other styles of writing. Moreover, the context and relationship of words within a sentence matters because the same words can be used in many different ways in any language. So if you want to write a conventional rule-based software, the number of permutations and combinations of all the rules to make sense of a single sentence or a paragraph will make it almost impossible for even the fastest supercomputers of today to produce results in a reasonable amount of time. Another important aspect of language is that it is always evolving, so these set of rules need to be constantly updated to keep up with that evolution of the language. This is where the conventional logic-based software breaks down!
Now if you consider the above example of natural language processing (NLP), even a toddler can do a better job compared to the fastest computers on the planet. The situation is similar if you consider some other capabilities like image recognition or speech recognition (a combination of NLP and sound recognition), you find that an even young child can do a much better job compared to the fastest supercomputers of today.
Now when you combine few such capabilities like for example image recognition and sound recognition, you have everything needed to create a completely self-driving car that never gets distracted with phone calls and texts, never get tired or falls asleep while driving and never drinks and drives! = Eliminating 93% of 37,000 deaths, 2.3M injuries and $230B in costs just in US alone!
Image/video recognition is also fundamental to having better robots doing almost everything physical that humans currently do including manufacturing, driving and cooking!
We talked about image/video recognition and natural language processing. But there are other things that human brain does better than conventional software including email spam detection, predicting the prices of flight/houses/stocks etc., recommending movies, music etc based on the viewing/listening patterns.
Hence comparing machines with human brain is not just a geeky obsession.
The implications of making machines equal or better than human brain, cannot be overstated.
Apart from being maybe a much better algorithm, there are 2 other very important and interesting features of the human brain that make it interesting for AI
A) It's ‘General Purposeness’: What we have observed in the brain is that the nerve cells in each region of the brain (performing different functions) are very similar to each other in terms of their structure and how they work. And given that we know our brain can learn and do an extremely wide variety of tasks, brain is clearly a computing device that has an architecture which is suited for general intelligence. This makes it absolutely interesting and potentially very rewarding because as it stands today, the “intelligence” or architecture(hardware+software) in today’s computers is definitely not general. Meaning that every time we want the machine to do a different type of task, we have to write a different software. Hence it's not so scalable.
B) Its Efficiency: Even if our brain consumes disproportionate amount of power (it consumes 20% of power with only 2–3 percent of body mass), it is still amazing that just a 3 pound object consuming less than 20W of power per day, can do things better than today’s supercomputers with massive amount of computing and power consumption. IBM’s Watson is a very good example of that. So we can learn immensely from the brain when it comes to efficiency (doing more with limited computing and energy resource).
