Various images for Sherpa
A big part of working with Sherpa was attending meetings and reading (research) documents.
I created a number of small images, and wanted share some here.
How AI works
At some point I created an internal document in which I re-wrote the Sherpa research results in an attempt to wrap it in a larger vision on society. One of the images I created for this document can be seen above. It tries to point out the issues that "AI" systems aren't clear-cut black boxes. They are complex "assemblages", which can combine cloud services, dubious data, and so much more.
The GPT-3 algorithm is at the cutting edge of AI.
Why tech should be regulated
I tried to visualize my thinking about thinking about technology.
I tried to point out how simpel models are often wrong. For example, technolical determinst thinking posits that the development of new technologies is the primary (soft determinism) or only (hard determinism) relevant factor in how society develops. This simplicity makes attractive and allows these thoughts to spread quickly. But because they are too simplistic, they can harm our ability to actually make good decisions.
I prefer more complex models that recognise that culture - people's ideas, our institutions and laws - also play an important role. If we don't recognise this, then we brush away our human agency. Silicon Valley told us "technology cannot be stopped", and because politicians actually believed this, it was allowed to become a self-fulfilling prophesy.
Kashmir Hill joked that How Normal Am I should come with a warning label that your beauty will be judged. So I created one.