Chapter 1 - Wittgenstein and Turing on AI: Myth Versus Reality

Summary

Amongst philosophers generally, the orthodox account of Wittgenstein and Turing on the mind appears to be that both were philosophical behaviourists. Amongst theorists sympathetic to Wittgenstein, however, there is an alternative – in its own way orthodox – account, which denies that Wittgenstein was a behaviourist (in any damaging sense). Proponents of this alternative account typically interpret Wittgenstein as opposed to Turing and as hostile to artificial intelligence – in their view, Wittgenstein belongs with famous AI naysayers. In this chapter, I shall argue that both the orthodox and the alternative account are mistaken.

A preliminary question, however: Is it perhaps anachronistic to discuss Wittgenstein's views on AI – a field frequently (though mistakenly) held to begin only in the mid-1950s, some years after Wittgenstein's death? To counter this concern, I begin with Wittgenstein's 1930s remarks on machines and the public discussion of artificial intelligence in the 1940s in the UK.

Human Computers

In Turing's groundbreaking 1936 paper, ‘On Computable Numbers, with an Application to the Entscheidungsproblem’, he set out the fundamental logical principles of the modern programmable general-purpose digital computer. Before Turing described his computing machines, however, we find Wittgenstein talking about ‘reading machines’ in his Blue Book lectures of 1933–34. Wittgenstein's usual example of a reading-machine is a pianola or player piano (Wittgenstein 1974: PG (henceforth PG), 69): he said that a pianola ‘translate[s] marks into sounds’ by ‘reading’ the pattern of perforations – which ‘we might call […] complex signs or sentences’ – in the piano roll (Wittgenstein 1972: PI (henceforth PI) §157; Wittgenstein 1965: BB (henceforth BB), 118, original emphasis). The pianola cannot do anything other than read; it cannot delete or modify the perforations on the pre-recorded roll. Nevertheless, a pianola can read ‘any pattern of perforations, of a particular kind, it is not built for one particular tune or set of tunes’, Wittgenstein said; in this sense, it is a general-purpose machine (BB, 118, original emphasis). Notably, reading machines may also include a lookup table: Wittgenstein remarked that ‘it is quite possible that there is a part of the mechanism which resembles a chart’ (in his example, the chart associates colour-words with colours) (PG, 190).