Computers are Tools for Humans

Watson competes on Jeopardy.I'd like to introduce a new theme that I will explore over the coming weeks: personificationism.  This first post will relate to personificationism in discussions of artifical intelligence; future posts will discuss notions of personificationism in theology, ecology, economics, and astrobiology.  

This idea grows from the Shinto wedding I attended two weekends ago.  I divided my discussion of the wedding into three posts: the first part was a brief analysis of Shinto; the second was a description of the procedural details of the traditional wedding ceremony; and the third discussed the very different procedure of the reception.  

As a rule, I try to avoid constructing meta-narratives of my own arguments as this can only limit what I hope is a broad and personally diverse set of interpretations, but the general theme of the first part is the nature and history of the received practices that we call Shinto.  The second part follows from this by describing a ritual that readers of this magazine should find utterly foreign and inexplicable but with which Japanese are intimately familiar (increasingly less so, but, as Kevin pointed out in the comments to the part one, Shinto being the wide base of Japanese culture explains a lot of what the outside observer might find uncanny about Japan).  The third part compares this foreign and inexplicable ritual to a more familiar one.  

When read in this light, the overall effect of the series should be to make the reader deeply self-conscious of elements of his own culture that he takes as true and objective properties of the world: these "true and objective" properties may seem just as uncanny to a Japanese person as a Shinto wedding would to a Westerner.    

Since this kind of cultural difference clearly exists, it follows that there is a set of received assumptions which are common to the human race and of which we are as unaware as the Japanese were unaware of Shinto until the importation of Buddhism from China.  The tendency to overlook these assumptions in argument has been called "humanism" by some and "chauvinism" by others; but both of these terms have alternative, unrelated, or confusing common usage.  I will refrain from discussing the general phenomenon and instead focus on a sub-set of this phenomenon which I will call "personificationism".

In the light of notions of personificationism, prevailing discussions of artificial intelligence are shown to be incoherent (like almost all mass-media science and technology reporting).  

Recently, Watson, an IBM-designed supercomputer, received much positive press coverage for defeating super humans Ken Jennings and Brad Rutter on Jeopardy.  It is this entire conceptualization that is problematic.  After the match, Jennings famously referenced the Simpsons: “I for one welcome our new computer overlords.”  Although this was a (pretty funny) joke, it reveals a major hole in received conceptualization of artifical intelligence.  The line should be more like, “I for one welcome faster and more efficient versions of our long-time computer slaves”.

We anthropomorphize machines both in popular culture and in media when really we should be seeing them as tools qualitatively no different than a hammer, a saw, or an Acheulian hand axe, albeit several orders of magnitude more sophisticated.  Most of the time this conceptual error is harmless, but in the world of scholarship such personificationism leads us to dead-end "ethical conundrums" like those implicated in Kurzweilianism and transhumanism.  Discussions of Facebook and Twitter revolutions are similarly confused.  Just as reason is the slave of the passions, so too will technology - no matter how sophisticated - forever be the slave of human desire.

As a corrolary: Watson and Deep Blue defeating exceptional human players at particular, predetermined games should not reflect the abilities of these machines-as-individual-entities so much as it should reflect the more-efficiently-applied aggregate knowledge base of a massive number of fairly intelligent human programmers for whom Watson is a tool.  The real meta-narrative in media coverage of advances in artificial intelligence should be that widely disseminated technology allows cooperatives of reasonably intelligent and motivated people to pool their resources and defeat previously untouchable intellectual outliers on specific, predetermined terms.