Least-Human Human

 

Earlier this year I read "The Most Human Human", a story of a real-life Turing Test AI competition, and the humans who competed. It's entertaining and thought provoking, but that's not my point.

It's fun to think about what makes a human "more human", but instead of the "most-human human", what about the "least-human human", or the "bottom-quartile-human human"? In the real world, human-ism is surely a statistical curve, and AI will pass the Turing Test against some individuals far sooner than others.

But more important is the Observer-Judge...they too must be facile and adept at what makes a human "human". Already we are each in this role today, against Twitter Bots and voice-assistant systems and photoshopped images and user interfaces which are some combination of AI and human designs. Is there a Turing Test for photographs? Videos? Language translators? Isn't the Turing Test a continuum of little machine/AI steps, against varying human opponents, with varying human judges?

I think already many are not equal to the task of being such a judge, and most us of will face challenges as fakes get better. At some point, should the Observer become a machine, when the average Turing AI is more-human than the average Observer-Judge, or average Human?

Yet we distrust AI doctors and Facebook fraud filters more than we do the human x-ray radiologist or the Russian story-bot. Why? Are we perhaps simply overly sure about our own level of humanity, collectively and individually, and thus our ability to be Observer-Judge or Human Participant in daily tests?

What fraction of people would say they are "more human" than average or median? What fraction would say they are more discerning about fakes and bots than average?

I think we can look around and see that the answer is "a large fraction", to the detriment of all of us. And, with improving AI, that fraction will rise. It's like a Dunning-Kruger curve; once you know the concept, you have to forever worry you're the blissfully unaware idiot....and yet those who do not themselves worry pose still more risk.

Comments

Post a Comment

Popular posts from this blog

Cars: How did our worst idea become our only idea?

  I saw a question on social media about "how did our worst idea become our only idea?" in response to car traffic. Surprisingly, I know the answer to this, both practically and theoretically.  Get comfortable and see what you think of my traffic manifesto: Network theory says that the value of a network grows as a function of the nodes connected, generally on the order of a square law (the factor is argued, but not the general statement).  This is known as Metcalfe's Law.  We already collectively knew this for millennia, which is why paths, streets, and roads have grown to connect everything related to humans, and you hear stuff like "all roads lead to Rome" and such.  The dawn of the telephone age formalized Metcalfe's Law, and the Internet has provided another data point, and recently Amazon and AirBnB have yet again.  With any "iteration with preferential selection" situation, you end up with winner-take-all patterns, as more users
Read more

Street Safety - Initiating Culture Changes

Image
  We previously outlined four major problems to solve to improve street safety, and two of these are related, as they both deal with responsibility. Figure out how to make somebody in relevant authority responsible AND accountable for street safety Alter our social and legal responsibility for crashes from individual obligation to collective responsibility. Let's dig in a bit. We know that changing broad social attitudes -- the culture -- is not easy. "Culture eats strategy for breakfast" said Peter Drucker. We also know how to actually change culture, and we also know that intent, discussion, convincing arguments, and cajoling will not do it. Changing culture requires forced action first, and then over time, with empirical experience, attitudes change. It takes pressure on a lot of implicit points, and it is generally a slow process. Yes, individuals can change their views on their own, but it's an even slower process with uncertain results. It's bette
Read more

Pragmatic Safe Streets Approach - Framing the Problem

Image
  First, let's double-check our framing context and be clear about goals. In my systems work, I try to always go up until the system I'm interested in, wherever I draw the line, and all of the systems it interfaces to, are in one mental picture. This frames the context for my system. Here, we said we were interested in street safety, which has so many connections it's hard to envision them all. Certainly it impacts all of the local city government, health and safety organizations, and street standards organization. Funding comes from the state and the Feds and gas taxes, so that pulls in still more. All of this is a bit of a red flag: when your system is "tightly coupled" with lots of hard-to-see interconnections, it will be tend to be hard to change, and unintended consequences can easily occur. Still, we need to make progress, so let's focus on the local city, and along the way be mindful of the tendrils that go further out, and see if we can prune so
Read more