[ExI] Popper and unscientific theories
col.hales at gmail.com
Fri Jun 10 01:57:33 UTC 2016
Here's another theory that sits astride this Popperian jibberjabber.
H1 = "There is no brain physics essential to the perfect functional
replication of a brain in any/all learning/knowledge contexts"
TestA_H1 = throw away all brain physics and compute models of brain
physics. Traditional computer substrates and neuromorphic non-von-Neumann
substrates .... all throw away the physics. The physics of the chip is the
physics of a model, not the physics in the brain that behaves according to
the model. I am not saying H1 is true or false. I am saying that the
testing is faulty. This has been going on in AI for nearly exactly 60 years
(60 year anniversary of Dartmouth in July!). Has H1 been 'proved' ? Nope.
Narrow AI every time. Yet H1 remains universally assumed by everyone over
generations to the point where, as science, AI is a Popperian miscreant of
the John Clark kind. H1 may be true! But the proper testing of H1 has not
begun. Consider the equivalent claim to H1 in another area: flight.
H2 = "There is no flight physics essential to the perfect functional
replication of flight (actual flying)"
But we tested this properly. Proved it false. If you delete all physics and
compute a model you get a flight simulator, not actual flight. If a flight
simulator is not actual flight, then why do we assume that a brain
simulator is a brain? Without any principle other than an intuition? And
then never actually test it! Is AI not AI at all, but merely a
computational study of intelligence in exactly the same way a flight
simulator is a computational study of flight? Well you'll never know if you
never actually test for essential physics, will you?
The 'science' of AI is unscientific in the sense Popper meant. But it's
not broken science because it's impossible to be a proper science. It's
merely a methodologically broken science because the proper testing never
gets done. I can see a TestB_H1 that fixes it. Maybe you can too. I have
been saying this since 2003. When is the penny gonna drop? How many billion
server farm clouds does it take to prove H1 either way?
The mother of all instances of popperian science cockups is that science
itself is not taught or understood properly or consistently. Ask 20 people
what science is and you'll get at least 20 feral and different answers. Why
is that? Don't we know?
On Fri, Jun 10, 2016 at 6:08 AM, John Clark <johnkclark at gmail.com> wrote:
> The philosopher Karl
> a theory is unscientific if it makes a prediction that can't
> regardless of how good experimenters become,
> but what
> a theory that makes lots of predictions that could have been proven false
> weren't and
> instead were
> the same theory
> also makes some predictions that can't be falsified?
> Should we just pretend those predictions don't exist?
> The Big Bang Theory makes a lot of predictions that have been confirmed
> and one of them is that the universe is 13.8 billion years old
> and so regardless of where we point out telescopes
> it predicts
> never see anything more distant than 13.8 billion years. And
> our telescopes
> have never
> n anything more distant than 13.8 billion years. T
> here are only 2 conclusion
> that can be
> 1) There
> lots of stars more distant than 13.8 billion
> years but we'll never be able to see
> because light hasn't had enough time to reach us and due to the
> accelerating universe there will never be enough time to reach us.
> Nothing exists that is more distant than 13.8 billion light years and t
> he Earth is at the center of the Universe.
> Despite what Popper might say I think #1 is the more scientific
> conclusion. In a similar way Everett's Many Worlds Theory does such a good
> job explaining how the 2 slit experiment
> I don't think it's unscientific to conclude other worlds might exist.
> John K Clark
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the extropy-chat