Frankenstein 2.0

For prior posts see blog 1, blog 2, blog 3.

For a blog post outlining the prototype ‘Dr Canonical’ and full details, see blog 4.

Frankenstein Download: Coming Soon.

This is a project I’ve been (belatedly) working on with Prof. Hayden Lorimer and Dr Pip Thornton (both at the University of Edinburgh) as part of a broader project on ‘Spoken Word Data‘ funded by the EPSRC Human Data Interaction (HDI) Network +.

As on the HDI website:

This project contributes to an ongoing academic and artistic intervention into the ways in which everyday language-use has become commodified by the technologies of digital capitalism, and the significant ethical implications this development is having upon our lives through personalisation and the transforming of the spaces in which we live. By finding ways to make the processes and politics of the digital economy more visible and legible to wide and diverse audiences, this project proposes two new artistic interventions which expose and challenge the ways in which AI based technologies such as search engines, home assistants, and other smart IoT devices transform, twist, and render into commodities the words that we speak

As part of the growth of home assistants like Siri and Alexa – and other voice-activated computation – we considered what it would mean to challenge and interrogate the promises of AI. In the case of Frankenstein 2.0, this was to offer ways to understand how machine learning – without big data and extensive (cloud) computing power provided by ‘big tech’ – could be reasonably worked with on a single desktop computer, by someone who is reasonably computer literate (i.e. me).

Thus, this project was born out of a condition of thinking what could be done, and how much could be gleaned, by feeding Frankenstein 2.0 with a variety of different texts in order to make it speak like an author. But, more importantly, what questions could we ask it?1 One of these is Mary Shelly and her work Frankenstein (and, of course, the inspiration for this piece of work). In future blog posts, I will articulate some of the technical detail and some of the hurdles of dealing with non-cloud based systems for voice recognition, for instance. The aim of this project is also to make it available in an executable format for others to try out.

The ‘prototype’ of this piece was on a plenary panel as Dr Canonical at the Critical Legal Conference – Frankenlaw – in Dundee in September 2021. This produced quite a few laughs, audio feedback (part of the valiant efforts at doing a blended, hybrid online/physical conferences), and some fascinating questions around the logics of AI for thinking about the law. This was a conference I’d never heard of before, but it was absolutely brilliant. Below are some tweets from the day.

1 Anyone who knows a bit more about this will realise that one cannot ask questions to a text without the machine learning having a range of question / answer data. Of course, we could not do this, so instead we simply use a Markov chain with a seed from a question.