On Freud and Autocorrect
Is an unfortunate but revealing auto-correction an instance of a Freudian slip?
Does the algorithm on my phone speak for my unconscious?
Perhaps an incorrect auto-correction even represents the Freudian slip in a purer and more demonstrative form. An incorrect auto-correction strikes us as coming from the outside, as the intervention of an agent who is not ourselves. And that is precisely the experience of the Freudian slip, of the unconscious interrupting the seamless flow of the operation of language in consciousness.
Auto-correct is an adaptive apparatus which deeply analyzes my words, and understands patterns in my use of language which I cannot articulate. In this way, auto-correct functions as a blind and dumb psychoanalyst. By completely removing itself from my person and my affect, this robotic therapist processes my language at a purely formal level, which ultimately reveals certain gaps and parapraxes.
The interesting thing about this algorithmic analyst is that it does not intervene to correct these parapraxes like a therapist, but actually creates a positive feedback loop where it attempts to reproduce them through guessing what I mean in moments of ambiguity. In this way, it naively surfaces the shifting signs and utterances of my unconscious by attempting to directly insert them into my discourse.
I wonder if deep learning offers the next frontier for psychoanalysis. By being able to analyze massive amounts of language at scale and in a purely formal manner, machine learning suggests new possibilities for analysts.