Bartlett Research Conversations: Jessica In

Learning to Joost: How Architectural thinking can enable extrapolative Artificial Intelligence for design

Speaker: Jessica In
Supervisors: Prof Penelope Haralambidou and Prof Sean Hanna
Guest panellist: Prof Roberto Bottazzi

Abstract

The term ‘joosting’, or ‘jumping out of the system’, was coined by Douglas Hofstadter as a measure of describing true creativity.[1] ‘Learning to Joost’ is therefore essential for any successful creative endeavour, be it a purely human, artificial, or synthesised hybrid of the two. 

This research explores a new framework for critically adopting Artificial Intelligence (AI) and Machine Learning (ML) as a methodology in architecture. By considering the nature of learning through different ML approaches alongside traditional acts of design, the research aims to establish a fresh approach to framing AI in architectural design.

The developing framework is put to the test through an experimental design exploration, a machinic reading of the Hwaseong Fortress Uigwe, Suwon, South Korea. Focusing on three key areas of architectural representation: the depiction of spatial depth, the reading of text, and semantics of visual classification, three traditional acts of design – observation, iteration and re-interpretation/remixing are reconsidered alongside a machine’s interpolative and/or extrapolative nature.

[1] Hofstadter, Gödel, Escher, Bach : An Eternal Golden Braid.

About The Bartlett Research Conversations

The Bartlett School of Architecture’s Research Conversations seminars comprise work-in-progress and upgrade presentations by students undertaking the Architectural Design MPhil/PhD and Architectural and Urban History and Theory MPhil/PhD. All current UCL staff and students are welcome to attend.

Held regularly throughout the academic year, the seminars are attended by the Programme Directors, Professor Sophia Psarra, Dr Nina Vollenbröker and Professor Nat Chard; PhD Coordinators, Dr Stamatis Zografos and Dr Stelios Giamarelos; and other PhD supervisors.

Image: Depth Map Reconstructions of Janganmun Gate; Jessica In