The BttT presentation was about using computing to design a better system to learn to identify plants. One of its goals was to become "self-obliviating technology", meaning that the more you use it, the *less* you need it and the less you depend on it. In this case: the more you learn, the less you need the system.
"Self-obliviating technology" is apparently an established concept, but we hadn't heard of it before.
It probably only applies to a small niche of problems, and of course current economic systems fear the spectre of degrowth, but even so we all kind of felt... I dunno... relieved? after that talk. It showed a practical example of how to make computing sustainable: by solving problems with it in such a way that we can get rid of it altogether in the long term.