23 October 2005

Arnold Kling on Kurzweil vs. Hawkins

I wrote a bit earlier about Ray Kurzweil and the Singularity, as well as about Jeff Hawkin’s On Intelligence. In this short article on Tech Central Station, I think Arnold Kling gets to the heart of the matter, and I agree with his conclusions about Kurzweil’s shortcomings.

Excerpt:
My sense is that Kurzweil basically thinks of the brain as disembodied. Although he frequently refers to our bodies, it is almost as an afterthought. In terms of an old mainframe computer, Kurzweil treats the body is if it were the punch-card reader, i.e., a rather quaint device for receiving input, but not nearly as significant as the Central Processing Unit.
Instead, after I read Jeff Hawkins (inventor of the Palm Pilot and author of On Intelligence), I became convinced that our bodies and our sensory experiences are an integral part of our intelligence. Kurzweil thinks of your brain as a computer programmed with a fancy pattern-recognition algorithm. Eventually, he predicts, scientists and engineers will “reverse engineer” this wonderful algorithm. On the other hand, I think that you have been exploring patterns ever since you played with your toes in the crib. It is this cumulative experience, rather than an algorithm, that constitutes your intelligence. The phrase “reverse engineer the brain” may sound plausible if one thinks of the brain as hardware plus software. But the phrase “reverse engineer your cumulative lifetime experience” may be more apt, and such wording carries with it no hint of plausibility.
No doubt Kurzweil is correct about the general technology pattern, but I think he goes far beyond what’s likely and reasonable.

29 September 2005

Thoughts on 'The Singularity'

The concept of ‘the Singularity’ has been in the air lately, probably due to Ray Kurzweil’s book, The Singularity is Here (I have not read the book, but I did read his earlier book). Here’s a quick summary of Kurzweil’s position on computers (by Kurzweil):
Just so that the record is straight, my view is that we will have the requisite hardware capability to emulate the human brain in a $1,000 of a computation (which won’t be organized in the rectangular forms we see today such as notebooks and palmtops, but rather embedded in our environment) by 2020. The software will take longer, to around 2030. The “singularity” has divergent definitions, but for our purposes here we can consider this to be a time when nonbiological forms of intelligence dominate purely biological forms, albeit being derivative of them. This takes us beyond 2030, to perhaps 2040 or 2050. [from this posting ]
What do I think about this? Some things come to mind.
Clearly computing technology is indeed increasing its power at exponential rates, and presumably this will continue. But brains are not computers, and we have very limited understanding of what we even mean when we use terms like ’smart’ and ‘intelligence’ and ‘consciousness’. I do believe that we will be able to create software that will enable much better pattern recognition (for example), probably on par with the abilities of humans in the next few decades. This type of thing would be very helpful, for example, in medicine when doing a diagnosis based on observed symptoms. Is this the ‘domination’ of non-biological intelligence over biological intelligence? Perhaps, but it seems to me that this will occur in certain especially computational areas. Somehow I can’t see salesmen being replaced by ‘non-biological sales units’ that negotiate complex deals (but will they get help from quick access to all sorts of information? sure). Certain areas of human experience are not very computational at heart – they’re about feel and judgment and creativity and learning. Technology may assist with certain of these areas, but won’t ‘dominate’.

Another Kurzweil quote: “I will point out that once we have achieved complete models of human intelligence, machines will be capable of combining the flexible, subtle, human levels of pattern recognition with the natural advantages of machine intelligence.” I would take exception to this, because I don’t believe we have a strong definition of what human intelligence is, and will perhaps always be expanding what we mean by it. I think it’s overstretching to just assume a ‘complete model’.

Kurzweil takes his position much further, into the realms of economics, medicine, etc. Read more here.

18 July 2005

'Visual Intelligence' by Donald Hoffman (1998)

Visual Intelligence is subtitled ‘How We Create What We See’ and in it Hoffman sets down a tentative set of ‘rules’ that humans use to interpret the visual input from our eyes. These rules work in combination to build up lines, shapes, colors, 3-d placement and more. You can try some of the visual illusions that help to show how the rules can allow us to be ‘tricked’ in many ways, such that your sense of what you’re seeing is not what is really there…

In the final chapters, Hoffman argues that we do the same thing with the input from our other senses as we do with the visual input; we create representations that are well-adapted to our survival. And he goes further, to say that our representations, while they must be taken seriously, are not necessarily a guide to what is ‘really’ out there…

I believe these are important points. Clearly we only pick up on certain spectrums of all that is ‘out there’ while other organisms are able to to pick up different spectrums, and may have very different (but also effective) rules for interpreting the input. Given this, can we really take seriously the idea that we are somehow close to understanding the universe? I’d say it would be foolish to think so.

13 March 2005

'A Different Universe' by Robert Laughlin

A Different Universe is very new book by Stanford physics professor (and Nobel winner) Robert Laughlin. His thesis is that we are leaving the age of reductionism and entering the age of emergence. By this he means that we may well have learned one set of fundamental laws of how matter works, but there are many limits to what we can do with these laws in terms of predicting higher-level collective phenomenon.

The book is written for the layman (ie. no equations!), and I still found some of the detail hard to understand, but the main point comes across. One example of what he’s describing is the behavior of elements as they go through phase transitions (while we’re used to thinking of gas, liquid, solid, there are apparently many other states at extreme temperatures). These phases exhibit certain properties which are consistent with the fundamental laws, yet have new behaviors that no one could predict from those laws. Only though experimental measurement have we found out about these states.  So the physical laws do not determine (and thus predict) all the outcomes, they are more like constraints.

I find myself in agreement with his thesis; I too believe there is much more that we can discover about how the universe works. The book is entertaining but perhaps a bit light once getting the main concept across.