The exact details of Adams behaviour will be published in Nature. I suggest we wait until this article is published before we try to determine how many angels can dance on the head of the pin, or what the limitations of my attempt to convey a simple point are, or what your limitations to understand that simple point are.
On Sun, Apr 5, 2009 at 5:54 PM, <[email protected]> wrote: > Imago 16 writes: > > " the FT article also makes a point of explaining what, exactly Adam did -- > produced a hypothesis, developed an experimental > methodology, and carried out an experiment. So it is in fact a big deal." > > I myself would not credit the FT with "explaining what exactly" the robot > did. For me, the most interesting suggestion about the robot's activity was > that > it "formed a hypothesis". I can vaguely conceive of a machine's "forming" a > proposition in the vacuous sense of, say, grinding out an infinite number > of > statements in the terms and rules of mathematical logic, but the machine's > being > able to distinguish or "recognize" a non-vacuous hypothesis is harder > since, > initially, any statement at all can be considered as a hypothesis. > Presumably > the scientists would get the robot around that problem by programming it > with > an arbitrary operational decision-procedure for classifying a statement as > a > "hypothesis". But the more programmed the robot is, the less > "independent" > it > feels, the more "mindlessly" mechanical it seems, the harder to credit with > "discovery" or even with "knowledge". But, granted, those are just > personal > reactions to word usages. > > FT said: > "Adam formed a hypothesis on the genetics of bakers' yeast and carried out > experiments to test its predictions, without intervention from its makers > at > Aberystwyth University. The result was a series of b simple but usefulb > discoveries, confirmed by human scientists, about the gene coding for yeast > enzymes." > > FT says the researchers also, "endowed Adam with a huge database of yeast > biology, automated hardware to carry out experiments, supplies of yeast > cells > and > lab chemicals, and powerful artificial intelligence software." > > From this, I can't tell if, as Imago claims, the robot "developed an > experimental methodology" or it simply proceeded according to instructions > from its > "powerful artificial intelligence software". > > That Adam mechanically detected and printed out genetic codes seems > unexciting today. Go to any DNA lab and you will see technicians having > lunch > in the > cafeteria while a machine back in the lab is mechanically doing such stuff. > Consider a coin-sorting machine of the kind you'd find in any bank branch. > To me > it feels silly to credit it with "discovering new scientific knowledge". > > Imagine that astro-scientists are given a new uniquely powerful telescopic > lens. They want to hitch it to machinery -- hardware and software -- that > will > keep it tracking the skies 24/7 to register and report hitherto "unseen" > celestial objects. They call their new robot 'Bozo'. > > To build Bozo, the scientists "endow" the telescope with a huge database of > info about the already "mapped" portion of the sky; they build automated > hardware that will mechanically drive the telescope's eye across the sky, > patch by > patch; and they write software that will compare the new scope's picture of > a > given patch with all previous pics of that patch -- and flag anything new. > (This is not unlike what the machinery at a medical lab does with your > blood > sample; the blood comes back with a "written" (machine-generated) report > citing > numbers outside a given range, etc.) > > From the FT story, it's not obvious how Bozo is essentially different from > Adam -- and neither of them, in this 21st century, seems anything more than > what > we'd expect coming up, given the technical developments of the past > decades. > > In fact, such sky-scanning "robots" have already been at it for some time > now. When FT says: > > "A laboratory robot called Adam has been hailed as the first machine in > history to have discovered new scientific knowledge independently of its > human > creators," questions have to arise in the mind of any alert reader. If you > read > anything about the Hubble telescope, you will find it credited with many > "discoveries". The FT piece tells us nothing that makes clear why Adam > should > be > thought of as "independent of human creators" but the Hubble or the little > floor-dusting "housekeeper" robots or even coin-sorting machines should > not. > > I personally think maybe crediting ANY machine with "DISCOVERING" things > may > be a bit much. But this may be fuddy-duddy of me, motivated by the > conviction > that no machine ever gets a "Eureka!" feeling. I admit we probably could > construct a stipulative definition that many of us would accept for the > sake > of > discussion. But, as always, stipulation is not creation. > > Imago16 writes: > > "Perhaps you should have read the article, Cheerskep, before commenting on > it." > > Many of us have been guilty of ad hominem jabs, Imago, and you and I both > recognize the dismissive insinuation in that line for what it is. But we've > been > trying lately to cut down on such stuff. Please help us. > > Not only that, but your assumption is wrong: For what it's worth, I report > that I did read the article before commenting. Go back and look at the > article > and my response, and note the difficulty with "discovery" and other > problems > was already there -- which it couldn't be without my reading the article. > > I still claim that Michael Brady's worry is misplaced, because no one will > ever be able to write programs that will distinguish future, unprecedented > arrangements -- of words, paint, musical notes, dance moves -- into those > that > will > be aesthetically pleasing and those that won't. No one can do this for all > of > us any of the time, or for any of us all of the time. > > > > > ************** > Worried about job security? Check out the 5 safest jobs in a > recession. > ( > http://jobs.aol.com/gallery/growing-job-industries?ncid=emlcntuscare00000003 > )
