Thursday, October 6, 2011

What Technology Still Can't Do (And Might Do At Some Point In The Future)

I wonder if there's a term for the set of experiences that are not shared, but that we know exist because we have experienced them ourselves. How do we know that another person experiences something called hunger? Because a) He tells us so & b) We have experienced it ourselves. Otherwise there is no proof of hunger.


There are so many of these experiences - dreams, pain, emotions, hunger, thirst, thoughts etc. Our understanding of these has been shaped by thousands of years of shared experiences. And yet we have never (in the literal sense) experienced someone else's pain or dreams or thoughts. We believe it when a person says hes in pain because we trust him. The level of trust will vary with, among other things, his physical condition. A wounded person for example, who says hes in pain is more likely to be trusted than a person who appears to be completely fine. But it still hinges essentially on our trust & judgement.

That's one real tough nut for technology to crack. Can it synthesize, and improve our sense of trust & judgement? Can you have a pain monitor to check how much pain a person is experiencing? Not in its current state.

Language is a beautiful thing. It has its set of rules, but if we so choose those rules can be modified to various degrees. Look at how much of material must have been written in English using jut 26 letters! Trillions of words! And increasingly, the written word is becoming automated. Search engines have become intelligent. You can have essay & story generators. These are all becoming increasingly sophisticated with greater processing power and software development.

However, technology still cannot generate essays ab-initio. The program needs repositories of words, it needs to be taught grammatical rules & syntax. But dont we humans learn the same way? I guess not. The base for computers was mathematics, and not language. The evolution of computers started with the diode and binary logic and Boolean algebra. Over the years, this is the paradigm which has been worked upon and developed. Thus, all computing technology based upon this will face inherent limitations because deep down, beneath all the advanced circuitry and LCD screens, it all still a series of zeros and ones. Its like a human being trying to learn English using Mathematics. Only the quantitative part remains, the qualitative gets missed out.

And this is true, not only for language, but for all art. Computing technology cannot synthesize art beyond a point so long as it is based on logic. I wonder what would have happened if the computer had been developed by the Medicine or Literature profession, instead of the Mathematicians, Physicists & Engineers. Medicine, Arts & Social Sciences are used to dealing with imprecision and subjectivity in their research and analysis to a far greater degree. What if the computer was not based on zeros and ones but on some other paradigm completely? Say a computer which would behave like a human, throw tantrums when overworked, respond with greater enthusiasm at times and so on.

Artificial intelligence will remain artificial because it is an abstraction, an oversimplification of human intelligence into only the quantitative part. Human intelligence is both quantitative and qualitative. (Diversion - to ensure your job is not made redundant by technology, try to work in an area which makes use of qualitative human intelligence) This explains, among other things, the failure of models to predict the 2008-09 Economic Crisis, even as some people were able to predict it.

To make artificial intelligence natural, it is necessary to rework from the grassroots. One can't synthesize the qualitative from the quantitative; A new technological paradigm from the ground up is needed - one which is free from the restrictions of binary logic. We need a system which incorporates qualitative-ness. And for doing that, the core computing technology will need to incorporate inputs from all spheres of human activity in its design. And who knows, perhaps, the next Siri will tell you what you dreamt of last night and help you in composing a blog on it...

2 comments:

  1. Human cloning and Stem Cell Research may take us there faster than one would probably expect

    ReplyDelete
  2. Absolutely possible...I for one would never underestimate technology :)

    ReplyDelete