What can histories of artificial intelligence teach us? On the development of large models and 'data-driven' research in AI

New

Join our Methods Fellow, Amira Moeding in a workshop which introduces methods of historical enquiry into the development of digital technologies and digital data. How can we do the history of technology today? What are the limits of historical enquiry; what are its strengths? Moreover, what can we learn from historical narratives about technologies? More concretely, what can the history of “Big Data” tell us about artificial intelligence today? What were, for example, seen as the pitfalls and problems with biases early on in the development of data-driven applications?

Together with you, Amira will think through and employ methods of historical enquiry and critical theory to gain a better understanding of the origin of ‘data-driven’ digital technologies. Therein, the workshop attempts to bring about both an understanding of the statistical or data-driven methods by asking how they came about and why they became attractive to whom. The workshop thus links technologies back to the interests and contexts that rendered them viable. This line of enquiry will allow us to ask what ‘technological progress’ currently is, how stories of ‘progress’ are narrated by industry actors, and what ‘risks’ become apparent from their perspective. By providing this contextualisation and recovering early interests that drove developments in artificial intelligence research and ‘Big Tech’, we will also see that progress, and the promises for the future that it holds, are not ‘objective’ or ‘necessary’ but localised in time and space. We will raise the question to what degree digital humanities cannot only use digital methods to aid the humanities, but how historical and philosophical methods can be employed to provide a basis for criticising and theorising ‘the digital’ and putting the methods so-called ‘artificial intelligences’ are based on into perspective.

If you would like to attend this course, please complete the pre-course questionnaire. We will prioritise places for students and staff in the schools of Arts & Humanities, Humanities & Social Sciences, libraries and museums. However, all are welcome to apply.

Number of sessions: 1

The course will provide a very brief introduction to historiographic methods. In the end. you will have an idea of how computational linguistics and natural language processing became, according to some, two different disciplines and why the basic methods that made applications such as Chat-GPT possible are still being criticised by, among others, Noam Chomsky. Therein, we will see how philosophical assumptions shape both hardware but also methods in artificial intelligence research and thereby challenge the often assumed 'neutrality' of technology.

Amira Moeding is a PhD student at Cambridge University in the Faculty of History, Amira’s PhD project focuses on the Intellectual History of ‘Big Data,’ how ‘Big Data’ as an approach to building artificial intelligence became thinkable, how data became foremost an economic resource, and how political imaginaries emerged from the possibilities associated with ‘data-driven’ technologies.

Amira came to Cambridge to study the MPhil in Political Thought and Intellectual History developing her interests in histories and philosophy of law and race. Before coming to Cambridge, Amira studied philosophy and cultural studies at Humboldt Universität zu Berlin, focusing on early critical theory and post-colonial critiques of private property. At Humboldt Universität, she also developed an interest in the philosophy of science (scientific models and representation) and mathematics (conceptions of proof).