Thursday, November 19, 2009

Cognitive Science, genetic engineering, iWife and 2012

Recently was discussing with my brother - what is the next big thing in the world of technology and we felt it was cognitive science. Earlier it was Internet and Computer Science - a wave that our parent's generations witnessed, next someone speculated genetic engineering - the progress there is slow and steady and once we make one big breakthrough then there will be no looking back. It can be used to solve lots of diseases and will revolutionize medical sciences, change our life spans. Also as my uncle says - we might start having genetically engineered children - which will initially face lots of moral opposition but over generations would gain acceptance much like (and much faster than) we started accepting gays.

The other big area where we see lots of work happening is cognitive science (cognition means "to know", common used form is recognition) where millions of dollars are being spend on making machine intelligent like humans. A very interesting quote I read somewhere that "An average general intelligence is much difficult to program than programming chess expert". It referred to the computer program that defeated the Chess champion Gary Kaspurov and having dabbled in machine learning myself, I can vouch for that statement. When I was in IIT doing my masters, in the second semester, majority of my courses were around machine learning - 3 out of 4 in the second semester- we had knowledge discovery and machine learning and computation brain. The small projects we did - like programming neural nets and genetic algorithms were very alluring and lots of fun. I still remember my favorite code being "Game of Kalah" that I did way back in my engg. days for a software competition where the computer could play smarter than me even though I wrote that code, the computer could beat me, the creator of the code.

I think in the area of cognitive science we have whetted our appetite enough and there are lots of people very very keen to know how much artificial intelligence will evolve in our lifetime. This is again something that will revolutionize our lives.

My short fiction( had a concept of iWife - customized wife coming from this fascination. And a number of people asked me - shouldn't there be iHusbands too :).

So on one hand we will have genetic engineering which will have customized human beings and on other hand we will have development in cognitive sciences where machines will be more free-spirited and human like.

It will be interesting to see who will end where. I hope we survive 2012 to witness this ;).


Siddhesh said...

Interesting, especially when you apply it to nanotechnology, and have millions of nanobeings getting more and more cognitive...

A la Prey

Scary thoughts, actually!

arati kadav said...

yes indeed ... also considering the climate change and all that - i think machines will more easily survive catastrophe than we humans would.

Abhishek said...

I have a few questions:
Firstly,as u said genetic engineering isn't moving at the the pace it should or could have. Now, extrapolating about ur own experience about making machines smarter than the creator, cant AI be used for quicker genetic engineering.
Secondly,shudnt we find niche areas where AI or genetic engineering for that matter be restricted to? For example, i certainly dont fear (or believe) any sci-fi stories about machines taking on man, what i do fear however is babies getting engineered (and we may not be all that far from that). That wud mean u can choose how ur offsprings will be, that of course if u have the money. Its like getting a car custom made if u throw in a few fat wads more!Scary for me, especially coz it will just increase our social disparity. Cudnt we just restrict it to ensuring our babies are born disease free, a much better prospect.

arati kadav said...

Yes Actually AI techniques are being used for genetic engineering very very actively.

Regarding restricting technology -that i think is impossible to do - it is only possible to delay. Where do you draw a line between tweaking a heart and tweaking a skin or tweaking brain? So it is ok for rich people to be able to live long enough and poor to die early.
Then what do you think about intelligence in general - what constitutes intelligence - most of us do not use 10% of brain fully. There are larger questions - how much of intelligence is diligence and common sense ? How much is good books and right mentoring during childhood, great circumstances? If you actually visit Einstein's house and see his documentary - you realize that he was the man blessed with great social skills surrounded by great people including his wife who did lot of his research for him.

Abhishek said...

That's interesting. Yes i agree, intelligence does have a large aspect which can be cultivated. Also, intelligence is probably wrongly used as a generic term, it has many aspects: memory, rationalising,produtiveness, grasp, sometimes even knowing how to cut corners! What is constant though about human intelligence as we know it that it has spurts of activity. Why a machine with even a fraction of human brain power (we shud have units for quantifying that!)will function much more effectively coz it wont have fatigue, distractions, insecurity, and all the complex aspects of thought. The challenge as i see it is to make a thinking machine without such frailties. Then again if a machine is working on an algorithm, could u call it 'thinking'?

arati kadav said...

That's the thing about neural networks- I have coded them and i don't know how it works ...There was a very interesting article I read about a difference between man and machine behavior. A man given a situation will react in some way and get rewards. Give him the same situation again - the probability of reacting in the same way is high but not 1. We are ambiguous by nature.

Algorithms are deterministic by design unless we start using random numbers and rand functions. But some psychologists say that we don't have rand functions - there are complex subconscious processes that determine our behavior. So again we face a contradiction. So the answer to your question is I don't know. I have programmed neural networks and I cannot tell how they work (something I have against neural network based codes). There are lots of trials and errors in it that again I hate. But then it works and that makes me wonder. Do they really approximate the neurons in my head. Can we have more such techniques. That's the belief on which hundreds of millions of dollars are being spent on. I can make a judgement one way or the other but the truth is I am not adequately informed as i am not standing in the boundary of that knowledge. I am not an expert :). But the experts believe they can. So there we are.