STU-1316

BBSMS | ML: OLD & NEW TYPES

Semester
Type
Option Studio
8 Units

Course Website

This studio has two objectives.

The first is to devise concrete proposals for an art-themed institution located in Manhattan, NYC. The second, to step back from site and brief and engage with the disciplinary questions of typology and design syntax.

We will take the long view and ask ourselves the following question: when it comes to designing, regardless of program or site, what are our options anyway?

While the 18th-century conceit that space ought to be planned according to some blueprint or type, is almost universally extinct, two recent unforeseen developments seem to have granted it a new lease of life. The first is the acceleration of overnight urbanization. In Asia, the prospect of housing 3,000 people into one building has rebooted the same historicist appeal to type which obsessed Europeans during their own post-war boom. The second is the black swan-type emergence of parametric design. Parametric design generates variation and versioning, creating new types along the way. These types are abstract and invisible; they do not recombine to generate variations but instead calibrate relationships, expressed in mathematical or computational terms.

We will explore the convergence of the “old” and “new” readings of design typology. Our purpose is to examine, through projects, the respective roles intuition and rules play in design.

BLOCK BLOB SLAB MAT SLAT | MACHINE LEARNING (BBSMS ML)
This is a software studio with attitude. The difference between intuitive and machinic does not mean doing something “manually” versus doing it “on the computer.” The machinic predates personal computing and dwells within any creative endeavor–architectural, literary, etc. –that is machinic in spirit. Conversely, there are intuitive patterns of machinic thinking, which creatives will not explain.

Drawing on the instructors’ expertise and deep skepticism of current paradigms of AI in design, the brief will incorporate aspects of Machine Learning (ML). ML is an analytical and generative framework for understanding multimodal signals such as images, audio, or space. ML provides a machinic approach distinct from traditional computational design strategies. With ML, design variations exist in a probabilistic space, rendering the design process one of discovery, rather than generation.

SCHEDULE
The studio will unfold in three phases.

The first four weeks will be devoted to the acquisition of core knowledge in parametric design, ML, and typology. The results will be evaluated on February 19th.

From then to the midterm (March 26th), participants will develop concept designs for an art-themed institution next to United Nations Headquarters in Manhattan. We will make use of multiple case studies of Modernist building typologies compiled by the cohort of 2015. The objective of the midterm is to select one concept design for further development.

Participants will then devote the remaining weeks to a scheme design for the selected option. The scope will be up to each individual participant. You may also choose to engage with questions of museography, drawing on instruments developed by Pan Michalatos for his class Quantitative Aesthetics.

All designs will be explored as both diagrams and physical models. There are no formal prerequisites for enrolment. A working knowledge of Rhino 8 /Grasshopper is a plus. The instructors will provide a bespoke environment suitable for ambitious and highly motivated beginners as well as sustained technical support.
A two-day site trip to NYC during studio travel week is planned.