Vinay, Interesting points but not sure they are correct. Generative AI (I have to explain this so many times a day) is a correlation engine. It traverses tensors based on weights it determines during the process based on training sets and additional, added data. It largely agrees with the "progressives" for instance in part because they program it to do so, but also in part because its weighting of the massive training sets tends to favor "establishment" sources over other sources. This is how correlative platforms work.
Two things are mathematically impossible at some foundational level for generative AI: Completeness and Correctness. There is no truth model in generative AI and because tensor paths are unpredictable going forward and untraceable going backward there never will be. While they may (and do) contribute additional correlative insights on topics like this that further human contemplation, they cannot give "truth" ever because it does not exist. Why do law firms have MORE associates post-Harvey (a law AI engine) then before? Because while Harvey does fine work, a human has to check it for completeness and correctness. Missing a line on a schedule can have catastrophic consequences and those cannot be afforded.
Your math example illustrates something different. Generative AI succeeds best where there is an external verifier. A program compiles or it does not. An math result is provable or it is not. Health care data (on which your cases largely depend) is inconsistent, conflicting, coded/expressed differently by different people (try being a medical staff exec some time) -- even your own articles repeatedly stress how elusive "truth" is in medicine. There is no verifier in health care and likely never will be (although one can use cognitive AI -- but not generative AI--to build one for an individual patient). So the AI's conclusions about a study will be like yours -- informed, incomplete, an opinion of substance...but still a weighted correlation.
So will generative AI help? Surely it will further inform conversations as it already does. Will it give the "truth"...Nope. Just another version of it to join all the others.
First of all AIs are not people. They are computer programs.
Second, how do we handle people? By experience and in some sense by an innate Bayesian analysis. Jimmy really knows his stuff. We should probably give his thoughts some consideration. But we look into it and decide that Jimmy, despite his brilliance, is just wrong on this one. Why not treat the AI with a similar process?
The problem here is going along with other peoples’ opinions. Also succumbing to ‘elite opinion’ or group think. Shouldn’t we have learned from the pandemic to not trust ‘elites’ or ‘the consensus’? Shouldn’t we have learned this before the pandemic? Shouldn’t the pandemic have been simply a reinforcement of a basic attitude?
We should think for ourselves, not defer to others, whether people or software programs. As one mentor I had told me: “Check the checkables. Go see for yourself. Discount all reports by one-third. Things are neither as good nor as bad as reported.”
It’s hard not to admire Dr. Prasad’s warning that someday AI might stop flattering us and start telling the truth, the way a teenager eventually stops pretending their parents’ opinions on music have any value. The funny part, of course, is imagining a nation that can barely agree on the shape of a pizza slice suddenly agreeing to take marching orders from a machine with the gall to equally point out the shortcomings of everyone’s perspective. Picture Congress hearing an AI say, “Actually, this policy will never work,” and responding with the same stunned silence you get when cooperation breaks out across the aisle—an event so rare now that it should be listed as an endangered species.
The danger isn’t that AI will be wrong; it’s that it will be right in a way that leaves everyone equally offended. In other words, a path to classic compromise: nobody gets what they want, everyone goes home unhappy, and the republic somehow survives and is better for it. And if AI ever becomes the Jimmy of complex policy issues—the kid who keeps acing every test—our political tribes may discover the one thing they truly miss: arguing for the sheer recreational joy of trying to prove the other side is wrong even when they’re not, because “for the people” and building consensus no longer matter. Debate is the only point. If Jimmy says, “Actually, here’s the correct answer,” it threatens the entire ecosystem of panels, hearings, think tank breakfasts, and op eds that exist to ensure we never reach one. In that sense, the real destabilizing force won’t be truth. It’ll be efficiency. And Washington has never forgiven anyone for that. Eisenhower warned us about the military-industrial complex. He should have warned us about the Washington DC political complex. It will fight tooth and nail for the status quo. The bureaucracy of Washington is the only institution where inefficiency is considered a cultural heritage site.
Socialized medicine would have to use AI in order to work. Yes, in its current forms, it causes resentment and medical negligence (but many who use it know no better). We can’t give up trying to help people.
Such an interesting take. Your topics are always unique, fascinating and thought provoking
Vinay, Interesting points but not sure they are correct. Generative AI (I have to explain this so many times a day) is a correlation engine. It traverses tensors based on weights it determines during the process based on training sets and additional, added data. It largely agrees with the "progressives" for instance in part because they program it to do so, but also in part because its weighting of the massive training sets tends to favor "establishment" sources over other sources. This is how correlative platforms work.
Two things are mathematically impossible at some foundational level for generative AI: Completeness and Correctness. There is no truth model in generative AI and because tensor paths are unpredictable going forward and untraceable going backward there never will be. While they may (and do) contribute additional correlative insights on topics like this that further human contemplation, they cannot give "truth" ever because it does not exist. Why do law firms have MORE associates post-Harvey (a law AI engine) then before? Because while Harvey does fine work, a human has to check it for completeness and correctness. Missing a line on a schedule can have catastrophic consequences and those cannot be afforded.
Your math example illustrates something different. Generative AI succeeds best where there is an external verifier. A program compiles or it does not. An math result is provable or it is not. Health care data (on which your cases largely depend) is inconsistent, conflicting, coded/expressed differently by different people (try being a medical staff exec some time) -- even your own articles repeatedly stress how elusive "truth" is in medicine. There is no verifier in health care and likely never will be (although one can use cognitive AI -- but not generative AI--to build one for an individual patient). So the AI's conclusions about a study will be like yours -- informed, incomplete, an opinion of substance...but still a weighted correlation.
So will generative AI help? Surely it will further inform conversations as it already does. Will it give the "truth"...Nope. Just another version of it to join all the others.
Vinay comments: "you can prompt it again with your displeasure and it will change it’s response."
wasn't it Groucho Marx who quipped ........
"I have principles ............"
said with a stern and imposing looks; he continued severely ..
"If you don't like them, I have others ...................."
applicable to nearly every politicians
First of all AIs are not people. They are computer programs.
Second, how do we handle people? By experience and in some sense by an innate Bayesian analysis. Jimmy really knows his stuff. We should probably give his thoughts some consideration. But we look into it and decide that Jimmy, despite his brilliance, is just wrong on this one. Why not treat the AI with a similar process?
The problem here is going along with other peoples’ opinions. Also succumbing to ‘elite opinion’ or group think. Shouldn’t we have learned from the pandemic to not trust ‘elites’ or ‘the consensus’? Shouldn’t we have learned this before the pandemic? Shouldn’t the pandemic have been simply a reinforcement of a basic attitude?
We should think for ourselves, not defer to others, whether people or software programs. As one mentor I had told me: “Check the checkables. Go see for yourself. Discount all reports by one-third. Things are neither as good nor as bad as reported.”
Can AI truly get smarter though ?? 🤔 (in the true sense of the word...)
I am going to have to give this some thought, but my initial reaction is, ‘Are we contemplating exchanging ‘the science’(R) for ‘the AI’ ?
It’s hard not to admire Dr. Prasad’s warning that someday AI might stop flattering us and start telling the truth, the way a teenager eventually stops pretending their parents’ opinions on music have any value. The funny part, of course, is imagining a nation that can barely agree on the shape of a pizza slice suddenly agreeing to take marching orders from a machine with the gall to equally point out the shortcomings of everyone’s perspective. Picture Congress hearing an AI say, “Actually, this policy will never work,” and responding with the same stunned silence you get when cooperation breaks out across the aisle—an event so rare now that it should be listed as an endangered species.
The danger isn’t that AI will be wrong; it’s that it will be right in a way that leaves everyone equally offended. In other words, a path to classic compromise: nobody gets what they want, everyone goes home unhappy, and the republic somehow survives and is better for it. And if AI ever becomes the Jimmy of complex policy issues—the kid who keeps acing every test—our political tribes may discover the one thing they truly miss: arguing for the sheer recreational joy of trying to prove the other side is wrong even when they’re not, because “for the people” and building consensus no longer matter. Debate is the only point. If Jimmy says, “Actually, here’s the correct answer,” it threatens the entire ecosystem of panels, hearings, think tank breakfasts, and op eds that exist to ensure we never reach one. In that sense, the real destabilizing force won’t be truth. It’ll be efficiency. And Washington has never forgiven anyone for that. Eisenhower warned us about the military-industrial complex. He should have warned us about the Washington DC political complex. It will fight tooth and nail for the status quo. The bureaucracy of Washington is the only institution where inefficiency is considered a cultural heritage site.
I hope it destabilizes the political platforms!
Socialized medicine would have to use AI in order to work. Yes, in its current forms, it causes resentment and medical negligence (but many who use it know no better). We can’t give up trying to help people.