Quite recently (yesterday, according to the time of writing this), I attended a rather interesting lecture on Cognitive Systems and Computing, by this dude named Ravi Seshadri from IBM. He spoke about the use of cognitive computing, or "thinking" computers, in analysing big data patterns and providing solutions in unstructured languages, like English. He went on a bit about Watson and its applications in computing, optimizing transportation systems, etc.
Now cognitive systems can allow us to interact with computers in natural languages, like English. Soon, you could probably tell your computer to make your coffee extra strong, and it would send the correct set of instructions to the coffee-maker to electronically brew your coffee just the way you like it. You could call out to your computer to book your air tickets, to wash your laundry, and so on and so forth.
Or can you?
Let's back this up a bit and see how we make the computer follow our instructions now. We see that we have to write our instructions to the computer in a painstaking, alien-like language that just barely resembles English. We give the label of 'high-level language' to those constructs and we call the people who can 'speak' this language as programmers. But there's a fundamental difference between this and our 'normal', unstructured languages: it's just that - they are structured. This means that we can give exact, precise instructions to the computer and the computer shall follow those orders to the letter - no more, no less, the ideal slave.
Now what happens when you introduce unstructured languages into the mix and develop on cognitive computing? The computer shall end up with the same fallibilities as that of a human. Develop the computer's database and information-acquiring algorithms enough, and what we've got on our hands is essentially another human, albeit one who can alculate really, really fast. Now this 'intelligent' computer will make the same decision errors as we do: when we can't decide what we want to eat for breakfast, how will the computer?
So the question is this: will using unstructured programming and natural language processing compromise on efficiency?
PS: I'm pretty sure that this entire post has been rather vague and hard-to-follow. But... that's the point. You see, I'm human - nowhere exact or precise ;)
Now cognitive systems can allow us to interact with computers in natural languages, like English. Soon, you could probably tell your computer to make your coffee extra strong, and it would send the correct set of instructions to the coffee-maker to electronically brew your coffee just the way you like it. You could call out to your computer to book your air tickets, to wash your laundry, and so on and so forth.
Or can you?
Let's back this up a bit and see how we make the computer follow our instructions now. We see that we have to write our instructions to the computer in a painstaking, alien-like language that just barely resembles English. We give the label of 'high-level language' to those constructs and we call the people who can 'speak' this language as programmers. But there's a fundamental difference between this and our 'normal', unstructured languages: it's just that - they are structured. This means that we can give exact, precise instructions to the computer and the computer shall follow those orders to the letter - no more, no less, the ideal slave.
Now what happens when you introduce unstructured languages into the mix and develop on cognitive computing? The computer shall end up with the same fallibilities as that of a human. Develop the computer's database and information-acquiring algorithms enough, and what we've got on our hands is essentially another human, albeit one who can alculate really, really fast. Now this 'intelligent' computer will make the same decision errors as we do: when we can't decide what we want to eat for breakfast, how will the computer?
So the question is this: will using unstructured programming and natural language processing compromise on efficiency?
PS: I'm pretty sure that this entire post has been rather vague and hard-to-follow. But... that's the point. You see, I'm human - nowhere exact or precise ;)
No comments:
Post a Comment