This post is a compiled snippet from a Twitter thread that I've transported over here for posterity. The original thread started here: https://twitter.com/b_cavello/status/1165323010229882880

Today, I was asked what I see as the biggest issues in AI.
Jokingly, I said there are too many. But I also wanted to think about it. Many of the challenges that face us are, of course, not AI challenges specifically, but are exacerbated or hastened by the technologies.

There are enormous philosophical questions around the inertia of governance structures and how swift and complete the enforcement of rules ought to be. The scale and speed of these technologies is in many ways unprecedented.

What would it be like if Dante was tasked with standardizing the Italian language in today’s world with predictive text autocompleting our phrases?
How might or could (or should) we harness such a power if at all?
This question is salient in modern translation and transcription.

To me, two of the most significant tensions brought to bear by automation are the following:
1) Agency: determining how much control one has over a decision or action
2) Consent: how much information one should reasonably have about expected outcomes of a decision/action

These may not be the best terms to describe these topics, and I welcome more appropriate alternatives if others are aware of or would like to create them.

“Agency” is significant for these complex systems like people or AI because we kinda build our societies around a belief that people can control themselves (and their tools, tbh)

It can be difficult to determine where the person ends and the tool begins.

“Consent” is significant in that the consequences both spatially and temporally of these technologies are far-reaching.

It is challenging to have informed consent in an environment with such complex interactions making the deep information retrieval capabilities dangerous.

It occurs to me that this tension represents a similar, perhaps grander tension between free will and determinism and the extent to which we incorporate or depend upon these concepts in governance.

Consent seems to be rooted in expectations about outcomes, perhaps not entirely determined but probabilistically comprehensible, at least.

Agency, on the other hand, seems to assume a relative imperviousness to conditions and influences making things relatively unpredictable.