Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
IBM’s Watson for President? Leftists Won’t Let Die Their Dream of Central Planning
There’s now a website, Twitter account, and Facebook page devoted to “advocacy of the artificial intelligence known as Watson to run for President of the United States of America.”
I mean, I get it. Wouldn’t it be great if our politicians could dispassionately review the evidence and pick which ideas and policies are best? And since humans can’t do that so well, obviously, maybe machine intelligence could. From the site:
Watson will be able to analyze trends in employment, markets, interest rates, education, poverty, crime, taxes, and policy to assess what actions are most suitable to accelerate investment in the nation’s future.
Of course this assumes the ethics and evidence and trade-offs for optimal public policy are clear. But that aside, it seems that the “Watson 2016 Foundation” really doesn’t need even a primitive AI since it has already chosen which policies are best, all by its carbon-based-life-form lonesome. Among the preferred polices of the “Watson 2016 Foundation”: “Single-payer national health care, free university level education, legalizing and regulating personal recreational drug use … shift bulk of electrical generation to solar, wind, hydroelectric, and wave farm … a minimum-wage that meets a reasonable cost of living.”
Hey, this almost sounds like the “Watson 2016 Foundation” pretty much cut-and-pasted the guts of the Bernie Sanders progressive policy agenda. (This does not surprise me since some leftists these days dreamily hope that AI can efficiently perform the sort of economic central planning that communist humans never quite could. And still can’t.)
What would be left for Watson do, exactly? Even weirder, the evidence supposedly supporting the above policies — such as “free college” for instance — is hardly clear. Watson may be pretty smart, but as for the “Watson 2016 Foundation” … I dunno. And just to push this concept a bit further, isn’t “Watson for President” really “IBM for President”?
Published in Science & Technology
Can Watson determine if the information fed to it is BS or not? Who gets to determine what information Watson gets to analyze before making its decision? I smell a mechanical CBO here.
Let’s see President Watson deal with a Watson from each of the 535 Congressional offices, and the ten thousands of Watsons burrowed into their government civil-service protected jobs. Or does President Watson just abolish Congress (b00) and the Washington administocracy (hurray)?
Sounds like a 1950s Isaac Asimov story. Sixty years ago this would have been science fiction. A stupid idea, but science fiction. Today it is just a stupid idea.
Seawriter
Why do we even pay attention to these idiots? Watson, has absolutely no intelligence. It is a glorified search engine and data base, how is that smart? That is like calling a camera a good painter.
All true, but that still makes Watson smarter than the typical Progressive.
Seawriter
Hasn’t any of these people watched Avengers: Age of Ultron and realized the kind of decisions a computer would make?
I’m joking, of course, but still…
Watson has it all over Hillary in the personal warmth department.
Right off the bat, the site unconstitutionally prejudges the parameters of Watson’s decision-making as President:
This statement presupposes that “accelerating investment in the nation’s future” is the mandate of a US President, but of course that mandate does not appear anywhere in the US Constitution.
Personally, if the parameters fed into the computer were strictly limited by the US Constitution, I would be very curious to see what conclusions the computer would come to.
Precisely, did anyone load the 9th and 10th amendments into Watson?
This idea didn’t fare so well the first time around when it was called the Romney Campaign.
Garbage in, garbage out.
Actually the evidence is in and it’s clear. Free stuff causes profoundly unwelcome unintended consequences in the medium and longer term; wage and price controls distort, retard, punish and harm even when they are as simple as minimum wages; more money doesn’t make education in general, better; stimulus doesn’t stimulate the economy beyond the specific pockets it reaches directly; easy money cannot predictably cause sustainable economic growth: central governments cannot centrally plan an economy for more than a few months, even when they use sectoral regulatory control as their planning devices; and most important for Watson, the information needed to use megadata programs is not knowable, and the information that is knowable is old aggregate data on which the most important variations have been averaged out.
Perfect SAT scores, no character, no common sense with absolute political authority and a fact set supplied by ideologues and charlatans. What could go wrong?
If only Pol Pot, Kim Il Sung and Fidel Castro had higher IQs and more data, what a wonderful world….
This suggestion betrays a failure to comprehend what Watson is and what it does.
Watson is impressive technology, but it’s not magic. To oversimplify a bit, it’s a combination of natural-language processing, linguistic analysis, and search-and-retrieval technology; it finds relationships in source documents, and it gives you results based on algorithms which rank the relevance of those results to queries you give it.
The results, depending on the specific application, might be a possible answer to a question, supported by passages from the source documents from which that answer was drawn; or they might just be a set of citations to source documents containing information that are relevant to your query, enabling you to research the matter further on your own.
Before it can do this, human engineers must spend the time required to curate the data, define dictionaries and type systems, and then perform extensive “training” (basically running test queries and then evaluating the accuracy of the results). Everything Watson “knows” is what you tell it.
Watson does not have opinions, and it doesn’t make value judgments. It is not smarter than humans; it’s just a tool for analyzing quantities of unstructured data that are impractical for humans to process. It can help humans to make decisions, but only by helping them find relevant data.
(Disclaimer: I work for IBM Watson, but I most emphatically speak for myself and not for IBM.)
I think Sanebox would yield a better result. Rather than try to make policy, just let the AI scan what comes out of Washington – putting all the junk into the trash.
I see: we’ve gotten to the point where somebody is talking about bringing Philip K. Dick’s short story “Stand-By” to life!
https://philipkdickreview.wordpress.com/2014/06/05/stand-by/
The premise is that the President is a computer, the Unicephalon 40-D, but there’s a human stand-by President who steps in only if the computer fails, which it never has. The stand-by president is a union job, given to a happily unemployed union member who’d much rather sit around and collect unemployment. One day, an alien invasion fleet is spotted, and they knock out the computer…
But more seriously, their premise is that being President is simply a technocratic job of optimizing various parameters, and there is, therefore, a “correct” policy on every issue. There is no role for politics in this view.
But as James points out, this website has picked its preferred outcomes. That is, they’re begging the question in favor of certain policies by choosing the questions they think are important. Other people might disagree with the list of important questions, and it’s even possible they’d disagree with the preferred outcomes.
As much as we roll our eyes at politics, though, that’s the role of politics. As I understand it, this was one of Niccolo Machiavelli’s contributions to political philosophy: the idea that government involves competing interests within society, and that there’s not necessarily a “right” answer. That’s not something we can ever hand off to a computer or a formula or a technocrat. The public simply has to hash it out for themselves in the political process.
I don’t think that’s necessarily true.
It could be more like the argument for self-driving cars that says they don’t have to be perfect as long as they’re better than humans.
It’s possible that computers are more capable of conforming to the limits of the US Constitution than human beings are.
A computer that answers “that is outside my jurisdiction as President” would already be performing better than the mean.
Who in their right minds would give control of a Nation to being with single digits worth of years of emotional development?
The United States elected Obama twice. He has the emotional development of a petulant 6-year-old.
Seawriter
Whatever that means.
I think they fail to understand one of the key insights of Hayek and modern systems theory: It’s not a problem of computing power: It’s a problem of information. No central planner can possibility run an economy efficiently, for the central planner simply doesn’t have access to the information required.
It’s not a matter of just collecting it – it’s the fact that the information about relative supply and demand requires near infinite amounts of local knowledge and also because absent a market process the information about the needs and capabilities of the public doesn’t even exist. That information is created through the process of bargaining and bidding, which forces us to prioritize our own needs and provides clarity in terms of how our own abilities stack up against others. Competition with others reveals information about the demand for goods and the capabilities of the marketplace. No central computer can know that through inspection, no matter how good it is.
Therefore, central planning is always doomed to fail, or at best be highly inefficient.
But at least it will be fair.