Hi Max, I would be curious to know how DMDU applied to AI governance contrasts with MTAIR (summary report). My initial guess was that DMDU is more promising in handling deep uncertainty and coming up with robust policy recommendations, but on second thought (my outsider’s impression of) MTAIR is also aiming at doing the same; I’m thinking in particular of this section:
Quantification and decision analysis: Our longer-term plan is to convert our hypothesis map into a quantitative model that can be used to calculate decision-relevant probability estimates. For example, a completed model could output a roughly estimated probability of transformative AI arriving by a given date, a given catastrophe scenario materializing, or a given approach successfully preventing a catastrophe.
The basic idea is to take any available data, along with probability estimates or structural beliefs elicited from relevant experts (which users can modify or replace with their own estimates as desired). Once this model is fully implemented, we can then calculate probability estimates for downstream nodes of interest via Monte Carlo, based either on a subset or a weighted average of expert opinions, or using specific claims about the structure or quantities of interest, or a combination of the above. Finally, even if the outputs are not accepted, we can use the indicative values as inputs for a variety of analysis tools or formal decision-making techniques. For example, we might consider the choice to pursue a given alignment strategy, and use the model as an aid to think about how the payoff of investments changes if we believe hardware progress will accelerate or if we presume that there is relatively more existential risk from nearer-term failures.
The reason I’m cautiously excited about your work so far (and this post in particular) is that, aside from MTAIR, it’s the closest thing I’ve found to sketching a potential answer to my question from awhile back, especially your discussion here, at least as I understand it (to be clear, I don’t work on any of this, so my question might simply be misguided—I’m simply interested in how systematic methods for decision analysis might help us make better decisions in high-stakes high-uncertainty complex systems scenarios).
Hi Max, I would be curious to know how DMDU applied to AI governance contrasts with MTAIR (summary report). My initial guess was that DMDU is more promising in handling deep uncertainty and coming up with robust policy recommendations, but on second thought (my outsider’s impression of) MTAIR is also aiming at doing the same; I’m thinking in particular of this section:
The reason I’m cautiously excited about your work so far (and this post in particular) is that, aside from MTAIR, it’s the closest thing I’ve found to sketching a potential answer to my question from awhile back, especially your discussion here, at least as I understand it (to be clear, I don’t work on any of this, so my question might simply be misguided—I’m simply interested in how systematic methods for decision analysis might help us make better decisions in high-stakes high-uncertainty complex systems scenarios).