A common issue I’ve found in AI Governance debates is that differences in opinions are often inscrutable. Small disparities in two people’s underlying assumptions can compound into major divergences in their predictions and ideal visions for governing AI optimally. I’ve found myself spending lots of time trying to understand the core points of contention between two perspectives so I can figure out where I land in the debate and which questions we’re really trying to answer.
To that end, I built a tool that uses the frameworks established IAPS’s paper Strategic Visions in AI Governance to break down where various thinkers land across core cruxes and which strategic vision they seem to advocate for most aggressively. The intention here is to accelerate the rate at which we can understand a given author’s world view so that those new to AI governance or unfamiliar with the author’s work can get up to speed quickly.
This tool does not perfectly represent anyone’s world view (I’m still tweaking the scoring pipeline to reduce inaccuracies), but I hope it can give a quick picture to anchor your priors—from there you can add nuance by reading their writing directly.
A tool for visualizing the worldviews of AI Governance writers.
tl;dr: I made a tool that uses the IAPS’s Strategic Visions in AI Governance paper to visualize different thinkers’ views on AI governance. Feedback is welcome!
A common issue I’ve found in AI Governance debates is that differences in opinions are often inscrutable. Small disparities in two people’s underlying assumptions can compound into major divergences in their predictions and ideal visions for governing AI optimally. I’ve found myself spending lots of time trying to understand the core points of contention between two perspectives so I can figure out where I land in the debate and which questions we’re really trying to answer.
To that end, I built a tool that uses the frameworks established IAPS’s paper Strategic Visions in AI Governance to break down where various thinkers land across core cruxes and which strategic vision they seem to advocate for most aggressively. The intention here is to accelerate the rate at which we can understand a given author’s world view so that those new to AI governance or unfamiliar with the author’s work can get up to speed quickly.
This tool does not perfectly represent anyone’s world view (I’m still tweaking the scoring pipeline to reduce inaccuracies), but I hope it can give a quick picture to anchor your priors—from there you can add nuance by reading their writing directly.
This is still an early version so I haven’t scored many writers yet, but I would love some feedback on whether anyone else finds this useful, if there are any features or writers you would like added, problems you want addressed, etc.