The subject of AI governance (managing artificial intelligence) has been gradually taking hold in my mind as something worth thinking about, something worth working on; a field where I might make myself useful.
80,000 Hours is a think tank that identifies and recommends working on the world's most pressing problems: areas where an additional contribution could make the biggest long term difference. Positively shaping the development of AI is one of these areas.
I'm a tech lawyer with a PhD from Oxford about the interaction between algorithmic curation of content and copyright law, and how this impacts democracy and individual wellbeing. I also have several years' experience in a contract automation startup. I'm no true coder but I know some simple commands on the command line of my MacBook, and I know the basics of git. (If only all lawyers understood the wonders of branched version control!)
My research and writing, so far, has been about copyright law, and how to align incentives in the online content sphere with the public interest. It's already clear that the operation of 'narrow AI' has produced new and difficult problems. The filtration algorithms behind search engine results and social media news feeds have contributed to filter bubbles and echo chambers; outrage-ification and polarisation of public conversation; and troubling hierarchies of power, influence and visibility.
There already an exists an alignment problem between our interests as individuals and citizens, and the operation of these narrow AIs that we have created to manage our information stream. Their operation is yet to align with our interests in a robust, diverse, humane, democratic information sphere. The scope of unforseen consequences resulting from the operation of these narrow AIs has been enormous.
How much larger must the scope of potential alignment problems be for broader, more general AI? If law may be effectively used to address the former (and I believe it can), then surely it has a role to play in dealing with the latter.
One of the disheartening things about working on copyright policy is the sense that the area is saturated, tendentious, stagnant. Debates go around and around without much impact on policy, and there are so many voices that the prospects of being heard and making a difference sometimes feel very small.
How encouraging, the prospect of sidestepping into a field where the opposite is true! I'll keep you posted on how it goes.
Comentários