Generative AI, Abstraction and Epistemology
The "Problem"
Coming up with the skeleton of a 50-60 minute presentation for technologists in the former CIS region about the topic.
What we did
There was a lot of brainstorming about the three topics and how they link together. A Keynote presentation was produced, and then four sheets of flip-chart paper. We iterated on the content several times, attempting to join it all together in a coherent whole.
The "Outcome"
We found that this is a rich field that may yield actually yield a paper. The lack of understanding about epistemology in the wider field is quite evident, and this informs poor choices about the value and content of generative AI. Since AI is seeking to "be" human and "do" human things, ideally better and/or faster than humans, but without a sound understanding of what it means to be human this is unlikely to succeed. AI is NOT human, it is a simulacrum of a part of human nature, that which can be subject to reduction. It is effective at the correct point of abstraction but it is without context outside of that point.