Wittgenstein, The Private Language Argument, and Artificial Intelligence
In his famous private language argument LW purports to show that no individual could have a private ‘inner’ language that was incoherent to any other. A private language is in fact impossible, [logically] so. In a much debated series of examples and arguments he comes to the ultimate conclusion that the only way we are capable of giving meaning to words is through an appeal to the practice and “customs” of our linguistic community. In other words, the meaning of a given word for a speaker of a language is determined by the pattern of its use in that speaker’s linguistic community, and a speaker uses a word incorrectly when her use is at variance with this pattern. In a different set of other words, there must be a public criterion of correctness. Therefore I could never use some term to refer to a private mental state because I would have no meaningful way to determine if the term was used correctly or not (since it is private there is no public criteria to appeal to). In my description of the argument I left out the part where appeals to inner criteria are ruled out, but suffice to say he makes a convincing case that such appeals are not possible. (see author’s postscript at the end for citation on the above.)
In my frequent writings on the topics of artificial intelligence I often suggest that other writers on the topic are guilty of a lack of precision in their language for using the term AI in grossly incorrect or incoherent ways. Mostly this is because, in my view, the only way to use it correctly is to use it in a description of the future. Though I personally find it highly unlikely within the next 500 years, AI could, at some point, become an actually existing thing. Other writers mostly disagree and commonly suggest that AI already exists. There are AIs driving cars, and playing Go, and running city’s traffic light systems among a host of other applications. The reason we disagree as to the existence of AI is, at root, because we disagree about the meaning of AI, it’s definition. My point in this post is not to quibble over the possibilities of AI but rather to try and elucidate my position with respect to the use and abuse of the term, and respond to a common counter argument.
One of the ways critics have responded to these attacks is by suggesting that the I have the wrong “definition” of AI. They believe the definition has changed from what is was way back in the 1950s when the concept was first conceived. They say that nowadays the “community” of computer scientists, technologists, digerati, digeratizzi , and others involved in the field have redefined AI to mean something totally different. Thus, I am wrong to suggest they are misusing the term as they are simply following the convention of the day. It would seem that Wittgenstein himself would support this position. After all, in the private language argument I just described, he appears to endorse the view that the only way terms are given meaning is through how they are used in the community of speakers of that term. In truth, these critics have a good point. It is correct to say that the “community” they speak of have redefined the term, they have given it a new meaning, one which is very at odds with its original conception. Moreover, Wittgenstein writes at length about the meanings given to technical terms and there too he suggests that they are given meaning by the persons who most use them, the community of scientists who write about and use the term(s) everyday in their work. As a Witt super fan I only very rarely take positions that are at odds with his own. In this case it would seem I must, as I just conceded that the critics points are accurate and in agreement with the core principles of the PLA, while my own position would seem to be at odds with it. I will try and show that one can still agree with the PLA and it’s conclusions regarding meaning, while at the same time disagreeing with the premise that AI has a “new” meaning today than it has historically.
I will do this by focusing specifically on one subtle point that was left out of my description of the PLA. This has to do with what precisely constitutes the “community” of speakers of the term, and how the makeup of that community and its ‘customs’ can in fact ‘corrupt’ meaning in ways that make a farce of the very concept of meaning itself. Here I will depart from Witt, or at least, depart from any heavy citation of, or reference to, his works. I do not want to defend my position on the basis of any philosopher, even one as great as LW, but rather on its own merits. Moreover I do not have the academic chops to pretend to be knowledgeable enough to use his words in such a manner. Therefore if the following section accords or conflicts with LW is of no interest to me though others are of course free to point out whatever similarities or differences they might find. That sure sounded like something LW would write didn’t it? I think in fact he did write something very similar (intro to Tractacus), lol!. How then can a community of users of a term ‘corrupt’ it’s meaning, and how does the makeup of that community influence it’s ‘corruptibility’. In the case of AI we have a term that is loaded with much historical baggage in the problem of the definition of intelligence itself. While one can certainly ‘look up’ a definition in the dictionary or many other sources they all tend to conflict in subtle and not so subtle ways. None agree with any other completely. It is fair to say their is no agreed upon definition of the term. Like pornography for the supreme court, one simply knows it when one sees it. In recent times and in the computer sciences and technology fields in particular their has been a tendency to parse intelligence into smaller and smaller ‘buckets” by inventing new terms like ‘special’ intelligence, ‘general’ intelligence, ‘deep’ intelligence, ‘systems’ of intelligence, etc. With each slice the concept becomes less and less clear and more and more confused. Wittgenstein himself was very much against the invention of technical terms that do not clarify and would no doubt be appalled by the practice but nevertheless it has occurred and the practice continues to this day.
to be continued…….sorry..ran out of steam..plus my wrists are killing me…..
Author’s note: I heavily cribbed from 1000wordphilosophy.com’s website and their very nice and compact description of the PLA. It is one of the most debated and controversial of LW’s many debated and controversial positions. It also happens to be one of the more convoluted from a man who’s straightforward is most other people’s impenetrable. In any event they do a nice job in one thousand words. I can say with almost certitude this would horrify Witt.