Retrenchment

From: Richard Loosemore (rpwl@lightlink.com)
Date: Thu Aug 18 2005 - 13:45:07 MDT


Folks,

I think there is a need for me to do some backfilling here, to clarify
some of the misunderstandings that lie at back of what I have been
talking about.

This backfilling is going to take some time, but I need to lay some
groundwork. Hence this message.

*Please do not take the comments below as arrogance*

AI research is, and has been for some time, in the most profound
trouble. I will try to explain what I think that trouble is.

There are several communities/paradigms/constituencies involved in this
domain. I want to quickly outline what those communities are, but I am
going to eschew labels for just a moment, because labels sometimes have
undesirable connotations. So here they are:

COMMUNITY (A)

Those who would understand completely if you started discussing
"Shiffrin and Schneider", or "Levels of Processing Theory", or "Deep
Dyslexia." They would also what a word-exhange error, a
morpheme-exchange error or a semantic substitution was. They would
understand the power-law of skill acquisition. They would know about
"Smith and Medin" if you asked them for their thoughts about classical,
prototype and feature theories of concepts, and what the current think
was on those issues, they might not be able to answer immediately, but
they would know what you were asking for, and would be able track it
down pretty quickly. And if you mentioned "motivation", "compulsion",
"obsession" or "pleasure" they would assume you were just using these as
shorthand for certain mechanisms, rather than assuming you were talking
dualist philosophy.

COMMUNITY (B)

Those who know how to write serious amounts of LISP, who understand
pruning algorithms in state-space search, and who know the difference
between Blackboards, GAs, and neural nets. They might well be able to
tell you about "unification" in Prolog, and be able to give you a
thoughtful discussion of the differences between John Koza's genetic
programming and genetic algorithms and evolutionary programming. They
would definitely know about goal hierarchies and planning systems.

COMMUNITY (C)

Those to whom the term "edge of chaos" is not just something they
learned from James Gleick. These people are comfortable with the idea
that mathematics is a fringe activity that goes on at the tractable edge
of a vast abyss of completely intractable systems and equations. When
they use the term "non-linear" they don't mean something that is not a
straight line, nor are they talking about finding tricks that yield
analytic solutions to certain nonlinear equations. They are equally
comfortable talking about a national economy and a brain as a "CAS" and
they can point to meaningful similarities in the behavior of these two
sorts of system. Almost all of these people are seriously well versed
in mathematics, but unlike the main body of mathematicians proper, they
understand the limitations of analytic attempts to characterize systems
in the real world.

COMMUNITY (D)

Those who have had the kind of experience in which they find themselves
fifty levels deep in a debugger, working on Final Candidate 7 of a piece
of software comprising 4000 source files of C and C++, in a hopeless
attempt to troubleshoot problems in a codebase that they have only
written tiny parts of, and in which the rest is mostly undocumented and
barely commented (by people who often did not have much English), with
the product due to ship in two weeks. An experience that bears about as
much relationship to a computer science degree as Mrs Featherstone's
Finishing School For Young Ladies does to a whorehouse.

COMMUNITY (E)

Those who could give you a reasonable account of where Penrose, Chalmers
and Dennett would stand with respect to one another. They could easily
distinguish the Hard Problem from other versions of the consciousness
issue, even if they might disagree with Chalmers about the conclusion to
be drawn. They know roughly what supervenience is. The could certainly
distinguish functionalism (various breeds thereof) from epiphenomenalism
and physicalism, and they could talk about what various camps thought
about the issues of dancing, inverted and absent qualia.

COMMUNITY (F)

Those who know enough about real neural hardware to think that there are
serious questions about whether the real computation takes place in
floods of junk in a synaptic cleft or in specific timings of incoming
spikes in the dendritic tree. They know about programmed cell death and
how that might relate to learning. They might know about the
Hodgkin-Huxley equations. They understand what Marr had to say about
the possible role of Purkinje cells in fine motor control, and they
would know way too much about the architectural features of the brain.

************************************************************

Now, some of these communities are more directly hands-on, while some
just watch and comment and contribute from afar, but the six different
language that they speak and the six different paradigms they bring to
the table are all in some way relevant to the task of understanding how
cognitive systems might work, and how we might go about building an AI.

But the problem is that you can go into one of these communities and
find very talented people who are completely ignorant of what is going
on in the others. Often, it is not just ignorance but actual scorn and
disdain, as if they are proud not to know what is happening elsewhere,
because they regard the ideas (and sometimes the people) in some of
those other communities as irrelevant or stupid. Frequently, people
have a smattering of some other field and think that they therefore know
it all.

A couple of examples:

I have watched a well-known cognitive scientist/psychologist give a talk
at a conference in which he drew a weird kind of neural net that he
claimed solved a big outsanding issue in the community. Other people in
the audience were lavish with their praise at the end of the talk, but I
had to stand up and try to explain to him that if we picked apart the
assumptions in his model, the hidden nodes in his net would have to
number approximately one for every distinct sequence of experiences in
the individual's lifetime, and each node would have to be totally
connected to every other. Does anyone here want to do a back of the
envelope calculation on that one? What happened, when I said this, was
that the guy could not understand what I was saying - he didn't try to
dispute my calculations, he did not *understand* how to derive the
number of neuron sets implied by his model. Nor did anyone else in the
audience, except for a couple of shy postgrads who talked about it
privately later. All the professionals in the audience thought I was
being an asshole.

I have listened to a well-known neuroscientist bombard his audience with
a 30-minute summary of the entire structure of the brain and what role
the various bits played in all aspects of cognition. Afterwards I tried
to ask him why, if we accepted his proposal that consciousness was
located in the LGN (or wherever he proposed to situate it that day),
this would have any bearing whatsoever on philosophical issues like
qualia, as opposed to just the mundane being-awake issues. Every one of
his responses indicated that he could not understand the difference
between qualia issues and being-awake issues (what is nowadays called
the Hard vs Non-Hard problems). He *did* claim to resolve the qualia
problem, it was just that he talked as if qualia were just a matter of
the system being awake, as opposed to asleep.

If there is anyone here on this list who can say that they have a good
grasp of all of the six categories above, could you please email me so I
can get a headcount? I assumed (foolishly, in retrospect) that most
would say yes. Now I think maybe not that many.

Could you consider the fact that, even if you *are* a person who is
fully conversant in all of these, it is not enough for there to be just
one such person working in isolation ..... that it should be an
absolutely basic requirement of anyone working in the AI field that they
can switch comfortably between all of these six paradigms and draw
meaningful connections between them?

Does anyone understand that so long as people live in these separate,
hermetically sealed intellectual compartments, the field of AI is going
to get nowhere?

To be sure, confined within a compartment, any given individual could
spend their entire life not challenging their own worldview, talking
only to people who agree with them and who are willing to tell them that
it is okay to refer to others as people who "just don't get it". But
then, strangely, that individual might get to the end of a long and
frustrating career and wonder why it was that that dream AI design
didn't quite ever seem to work, although it always seemed so close...

I have seen closed-minded cognitive scientists who think AI people are
just engineers who play with fantasy toys, and who are willing to call
any paltry little algorithm "intelligent".

I have seen closed-minded AI people who think that cognitive scientists
just fantasize about how the mind works, but never try to actually build
anything that works.

I have seen closed-minded philosophers who invent computationally
idiotic thought-experiments and then claim to be able to prove that AI
people are idiots.

I have seen closed-minded software engineers who think that AI people
just play with toy systems and don't know dick about building real
oftware - and who also think "psychology" is what shrinks do.

I have seen closed-minded neuroscientists trying to tell cognitive
scientists that their fMRI machine has found the part of the brain that
is responsible for sarcasm, and then dismiss the cognitive scientists as
fools when they say "Oh really?".

But then, by contrast, I have to say that most of the Complex Systems
folks are extremely accepting and rarely closed-minded. But they tend
to spend their time making connections between different fields rather
than settling down to apply their skills in one area, so they tend not
to acquire depth in the other five of the above fields.

CONCLUSION

We need a revolution here, folks.

We need people from the various communities who are willing to get
together and seriously educate themselves in the other fields without
feeling like this is some threat to their egos.

Or, failing that, we need those who can't change to step aside and let a
new, unpolluted-by-prejudice generation step in and start getting some
real AI-construction done.

This is a prelude to amore lengthy work, so in due course an expanded
version will be available. For anyone that cares.

Meanhile, I don't want to debate the above in this forum. You can't
debate issues that touch people's core paradigmatic commitments. Read
Kuhn or Feyerabend to understand why.

Richard Loosemore.



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:51 MDT