Bit of a digression but I think you might be too optimistic when you write "When you’re communicating with stakeholders, ask: Are we using “probabilistic” to mean the same thing?".
Anecdote time: I once had to explain to a fairly senior executive why the average of the entire population wasn't just the average of the sub-group averages. And when he did finally get it, he still said something like well - that's all well and good but all of us "know" the number should really be X, your data must be wrong.
The craziest thing was that wasn't even my main point - I wanted to show that the population had a bi-modal distribution so taking a "point estimate" to feed into a financial calculation that was going to be used to set policies - regardless of whether it was a median or the mean, was not going to be "representative" and a scenario based approach made more sense.
Haha. Agree and know what you mean. I think that's just the way the world is.
I used to do Basel capital policy. And one fine day, I realized this super senior person in the same domain did not even understand the basics of how capital arose from a balance sheet. I think we just have to assume the lowest common denominator all the time. :)
Couldn't agree more, what if misinterpretations of core terms like 'determinism' - especially the precise definiton Thinking Machines presented - lead to a muddled understanding of actual AI capabilites, resulting in ineffective governance frameworks or even misplaced public trust?
Bit of a digression but I think you might be too optimistic when you write "When you’re communicating with stakeholders, ask: Are we using “probabilistic” to mean the same thing?".
Anecdote time: I once had to explain to a fairly senior executive why the average of the entire population wasn't just the average of the sub-group averages. And when he did finally get it, he still said something like well - that's all well and good but all of us "know" the number should really be X, your data must be wrong.
The craziest thing was that wasn't even my main point - I wanted to show that the population had a bi-modal distribution so taking a "point estimate" to feed into a financial calculation that was going to be used to set policies - regardless of whether it was a median or the mean, was not going to be "representative" and a scenario based approach made more sense.
Haha. Agree and know what you mean. I think that's just the way the world is.
I used to do Basel capital policy. And one fine day, I realized this super senior person in the same domain did not even understand the basics of how capital arose from a balance sheet. I think we just have to assume the lowest common denominator all the time. :)
Couldn't agree more, what if misinterpretations of core terms like 'determinism' - especially the precise definiton Thinking Machines presented - lead to a muddled understanding of actual AI capabilites, resulting in ineffective governance frameworks or even misplaced public trust?