Wednesday, April 6, 2016

Oh, You Thought Robo Was Just a Japanese Sex Doll

Artificial Intelligence is a lot closer than those twisted Japanese sex dolls they make now.  Unknown what a plastic blow-up sex doll cost when you could buy one from the back pages of "Popular Mechanics" but we're guessing these may cost a wee bit more.

But you're looking for more than porno, right?

Well, you are, aren't you??

OK, then ... AI approaches, slowly and with a sly promise in its eyes, and you feel that familiar stirring within you as ...

(Ed:  focus, please)

Yah, so AI approaches in subtle ways and the latest is in a tool to assist radiologists in their diagnostics.  There is software which will analyze X-rays, CT scans, and MRI scans thus 'relieving diagnosticians' of some of the more burdensome aspects of their work.  (ScienceDaily:  'Deep Learning' used to assist overburdened diagnosticians)

There is clear merit to reducing the workload of overtaxed staff but doing it in this way supplants human interpretation and thus robotizes something beyond a mechanical skill.  Building cars with robots is one thing since it follows a memorized process in the same way every time.  However, something requiring active thought does, in its purest sense, still follow a memorized process, since it's at root a computer program, but the process has so many more variables, logic paths and general complexity.



In another twisted example of Artificial Intelligence, a program put together a compilation of all of the works of Rembrandt and came up with its 'vision' of what another painting from him would look like.  (The Guardian:  'New Rembrandt' to be unveiled in Amsterdam)

Note:  it looks like all the others.  Not a huge fan of Rembrandt.

Even at its best, this type of AI isn't even a flyspeck on a real artist's backside since its only manifestation is in digital art which has no idea of the feel of the brush on the canvas or any of the myriad variables flying about in the artist's crazed mind.

Robo Rembrandt gets away with it because Rembrandt did so many portraits.  The only way to get beyond that is to review the works of, say, van Gogh to analyze the general content, the brushstrokes, the style, etc.  If Robo Rembrandt can do that and produce something unique which we would believe was done by van Gogh then we may consider it to have some artistic merit but short of that it's generally unremarkable.


Simulation of human thought in this way is another manifestation of the same thing as with interpretation of radiological information.

Robo has gone way, way past Japanese sex dolls ... but ... here's one now just because we love you.




Take it easy, cabron.  That's a doll.  Why it has the face of a child and boobs like Mount Vesuvius is for the interested student to discover.  We don't want to know.  (No links as you can find your own.  There are many links on Google so you won't need "Popular Mechanics" anymore.  Unsure if this was one of the models for $8K or so.  No, I'm not kidding.)


The advancing AI is much more subtle than M.I.T. Nutboy who made the simulacrum of his favorite movie star.  That's just another variation on the theme of Japanese sex dolls whereas the diagnostic software, etc listed above goes into significant areas of human thought and seems representative of some form of evolution since we're sure you can think of other examples as well.  Self-driving cars give us another one.

The concerns in this regard are not simple Luddism as we will really be diggin' it when NASA finally gets around to building a starship but some aspects of hardware / software / medical evolution are disturbing and we're not so much concerned by the facts of them but rather what extrapolations can be made based on those facts.

There was an article earlier about inductive reasoning and there was a question about whether that starts with a conclusion and, in effect, works backward but that's not the case.  Here it applies in that we have the facts of these examples of Artificial Intelligence so what conclusion is permitted based on that knowledge.  The Rockhouse judgment is it's still too early to know and we need more facts but we definitely see reason to continue observation.  (Blog:  Inductive Reasoning and Anti-Vaxxing the Bubble Queen)

No comments: