Colloqium #32
19 October 2024
Description
What do we mean when we talk about ‘AI’? To whom does this field of research belong? What do LLMs have in common with demons and improvisational theatre?
These questions, one might think, are trivial; with some going to far as to say stupid. But unfortunately artificial intelligence (AI) must be considered an ‘unscoped system’, as Dr Timnit Gebru observes in a lecture (check video in the links-section), citing systems engineering (link below). The conclusion is that ‘AI‘ has the goal of ‘doing everything for everyone in every environment.’
So what does it mean when it is claimed that AI has solved the problem of protein structure prediction? (link below, see also: Nobel Prize in Chemistry 2024)
Why did some people believe LaMDA 2, a google-internal LLM c.a. 2022, achieved self-awareness? (link below)
Why does AI simplify the work of scientific/academic practice, but complicates doing good science?
I have many questions, a few (incomplete) answers and am looking forward to an exciting discussion.
Speakers
Jan Krämer: Jan Krämer is an ordinarily odd computer scientist, who, after a detour through cardiovascular physics, is now engaged in the AI-SKILLS project at Humboldt-University Berlin. The project’s objective is to assist lecturers across all faculties in engaging students with AI methods and technologies in a subject-oriented manner and to facilitate the successful integration of these into the higher education curriculum.