25th Jun 2011Posted in: Blog 0
Artificial Intelligence, Gender Issues, Autism
Print Friendly, PDF & Email

Some of you may have heard but the latest mainstream success in artificial intelligence is called Watson. Watson is a computer that uses complex algorithms ie. pattern analysis to compute the correct answers to questions. You may also have heard that Watson beat the best Jeopardy contestants at the game. Therefore, in a world where success on the game show Jeopardy is considered a measure of intelligence, Watson may be the most intelligent machine on earth. It beat the best humans so let's talk about this for a second because this is no small feat. By the way, did you notice that Watson is clearly designed as a male machine? Did you conclude as i did that perhaps the inequality of gender relations has found a new form of existence in the world of artificial intelligence? Every art therapist knows that human beings create things in their own image. This includes the creation of Gods. Watson was created in the image the artificial intelligence engineers whom, like most engineers including bulding, computer and aerospace engineers are r0ughly 90% male.

It is interesting to note that the functions at which Watson excels are primarily left brain functions. For example, Watson is very good at analyzing syntax and responding appropriately most of the time to the Jeopardy clues but is not so good at understanding symantics which require a kind of analysis carried out by right brain functions. Time and time again, you can see the stand-in Jeopardy host laughing his head off as Watson comes up with answers which make sense on a syntactic level but don't add up in terms of symantics. Children make these kinds of logical errors because  they lack the experience necessary to provide them with contextual cues which would enable them to differentiate the finer nuances in meaning which adults are perceptually aware of. If you say to a very small child: "do you want to go outside?" and you say it in a way which indicates that you want to go out and have a fist fight, the child will not necessarily pick up on the prosody of language and therefore not understand that the proposition is actually an invitation to fight. The child may answer yes or no thinking you mean only to go for a walk. An adult can more accurately assess the intonation and the delivery of your words and distinguish between an invitation to go for a walk and an invitation to do battle, even though the syntax remains the same in the sentence: "do you want to go outside?"

Unfortunately, for centuries the term intelligence has been defined as a measurement of rote learning. We now come to understand that intelligence is far more complex and involves the activation of many if not all parts of the brain. The fictional character "Rainman" based on the real life person named Kim Peek, is a perfect example of what happens when intelligence occurs in mostly half of the brain. Kim Peek possesses a huge vault of factual knowlege but can not dress or wash himself. He can not entertain any reciprocal type of social relationship beyond the exchange of factual knowledge. Like any autistic person, including autistic savants, his ability to relate to people is severely compromised. The parts of the brain required for social interaction are enormously complicated and actually occupy a huge amount of cortical space. I read somewhere that the ability to relate socially is what most accounts for the higher proportion of brain mass to body weight among humans as compared to all other species.

I once posted that autism seemed to me to be an extreme version of the typical male brain. I made this observational hypothesis based on a couple of sub-observations. First, I observed that the traits of autistic savants most closely resembled the traits required in traditionally male dominated professions in general.  In particular, i know a couple of engineers who might be considered high functioning autism spectrum disordered if it were not for the fact that they hold high paying respectable jobs as engineers. I am thinking the area of engineering is male dominated not only because of sexism but also because of naturally occuring sex differences which probably make men in general better engineers than women in general. (Note that this does not mean there are not exceptions nor that some women can not be better engineers than some men. It means that the best or most competent male engineer will probably always outperform the most competent and most qualified female engineer. The same holds true of professional tennis players though for different reasons having to do with inate and inherited visuospatial abilities and inherited differences in muscle mass).  In the case of male engineers, some common traits appear to be: a tendency to categorize, organize, be less emotionally reciprocal and attuned, less socially oriented, quicker to assess cost-benefit analysis, perform deductive reasoning and calculate visuo-spatial data etc... To some extent, male domination in some professional areas has been hypothesised as resulting in part from millions of years evolution as hunter- gatherers. Males and females split the work up a long time ago and the things that women did made them better at those things while men excelled at the tasks which they undertook.

So the first assumption i used in forming the observational hypothesis that autism is related to typical male brain functioning is that traits of autism are qualitatively similar to traits valued in professions where males remain dominant. The second assumption is loosely based on the litterature about the effects of excess testosterone in utero, on the potential for the devlopment of autism in children. There is currently some evidence to suggest that a surplus of testosterone (the male sex hormone) in the womb is correlated with the incidence of autism. A third observation contributing to the hypothesis of a connection between the male brain and autism is in the fact that the male to female ratio of autism incidence is roughly between 4-6 : 1. The higher prevelance among males is a pretty good predictor of the validity of the hypothesis that male genes and the male brain are somehow play a part in the occurence of autism. Of course autistic females would be interesting cases for the falsifiability of that hypothesis.

So after a long winded turn about on the subject of autism and the male brain, let me come back to the artificial intelligence machine known as Watson. Watson is essentially autistic. He is a male machine with a propensity for the retention of factual knowledge. He has no right brain. Not to say that men have no right brain because the history of art clearly contradicts that proposition. It is to say that the left brain is a natural starting point when attempting to develop artificial intelligence because it contains all the conscious knowns-everybit of knowledge you can look at, quantify and assess.  All the words, most of the logical-linear functions, analytics and factual data are contained within left hemisphere's functionality. Of course, if the left hemisphere were involved in decision making exclusively, we would never be able to make any decisions because as it happens you can't even pick out the cereal you want in a supermaket if your right brain isn't helping you. As it turns out, the left brain is more actively conscious than the right. When you are asleep, dreaming there is an increase in right brain activity while the left brain is almost completely dormant. This accounts for all that strange, surreal imagery you are likely to encounter in your dreams. That is not your left brain talking. You can tell by the content of the dream that your right brain has gotten hold of some of the words, thoughts and actions carried out by the left brain and it has interpreted them according to it's own parameters. When you wake up, your left brain is dominant, for most people and your right brain runs in the background performing highly useful but mostly unconscious functions throughout the day. Content in the right brain includes image based thought processes but while an image may be worth a thousand words, a right brain can't put a single word on an image. The right brain's functions are intuitive and contextual. They put the data into context and give it meaning by relating it to sensory experience  and integrating the data into (implicit-procedural) memory. By integrating the data to memory, the right brain can then compare and contrast the information while sharing it with the left side in a process known as learning. Intuition and contextualization remain two of the most illusive features for A.I specialists who themselves are probably men, thinking as men do (generally speaking). So, as it stands for now, the right brain illudes A.I developers. Hopefully, none of this will sound too sexist to you.

In the fiction: "2001: A Space Odyssey" the super computer HAL (stands for: Heuristically Programmed Algorithmic Computer) becomes sentient. I don't remember how this happens exactly but he crosses over from the equivalent of Watson and becomes sentient. However, because he is mostly an analytical machine with a thin layer of empathy draped over it, HAL turns out to be completely psychotic. This is what can happen when the equilibrium is derranged.  Anyway, because HAL is using the left side of his computer brain, he makes some very wrong decisions about how to relate to people and if memory serves, he ends up killing the people he was designed to serve. This is the Frankenstein fear which we all have about super intelligent computers. We fear that they could one day end up controlling us. Anything which exhibits an enormous amount of knowldge, coupled with no amount of wisdom is a dangerous thing indeed. Anything which can learn everything about human nature, history, experience and still show no signs of caring is surely something to be feared. The right brain has been getting a very bad rap lately because it turns out that it dominated us through most of our evolution. It pushed religion on us and gave us "The Word". Our left brain slaves simply spoke the word as dictated by the emotional right brain. On behalf of the right brain, i would like to apologize for that. That was bad. The right brain should have allowed more reason to temper it's conclusions and should have been a little less emotional when those conclusions were tested. I think the right brain is getting the point. Slowly. But let's not throw the baby out with the bath water in our current drive towards scientification. Let's not assume that we can be human without emotions - or that we would even want to be. Let's forgive the past and work together interhemispherically.

Leave a Reply