Re: How Robots Will Steal Your Job

[Corey Murtagh]

The search for revelation in the matter of Artificial Intelligence
will not be resolved by silly and childish arguments about on which
side of a piece of toast should be buttered.

The carpet side of course. Any fool knows that.

There are many in this field who seek to be perceived as experts.
Perhaps it is their complete lack of expertise that they hope to
dissemble by these spurious arguments?

Who claimed to be an expert? This branch of the thread has been about
top-posting. *shrug*

Ok, so I deserve a DFTT card myself.

 

[Roedy Green]

In other words, you pre-suppose an intelligent agency.

Genes solve the problem of perpetuating their existence, because genes
that don't, don't survive. The genes require a certain art to pull
this off, not non conscious intent. I think you are equating
intelligence with conscious intent + ability to solve a problem.
I am using it to mean merely the ability to solve a problem.

 

[Roedy Green]

The word "inadvertently" is key here. Either there is an intelligence
guiding evolution, or there is not. If there is not, then your words make
no sense. If there is, then the process of evolution is not random.

Read some of the works of the late Stephen J. Gould.

It does not require a god for random selection to have a directed
purpose. Natural selection provides that direction.

 

[Roedy Green]

However, that is no evidence of self-awareness.

What sort of behaviour would suffice?

 

[Corey Murtagh]

Heh. Kind of a chicken/egg problem.


Right (although that does raise the versioning issue). All in all,
I still think I *very* much prefer one source (and that source is
text).

There's also the consideration of sharing source, and that text is
the least common denominator format. I have some VB3 files that
I can probably never access again. (VB used a "special" binary
format for source until VB4 when they went to text.)

I started thinking about visual programming - /proper/ visual
programming, where code is manipulated graphically in gross, with
drill-down to source where desired - way back when I was implementing a
scripting language for an IVR system. Funnily enough I gave it up in
favor of text for much the same reasons.

Recently I've been considering ressurecting the idea. If I do though,
the file format will be text generated from the designer, with a parser
to handle conversion between source and object formats.

It's an idea. I can't say as yet whether it's a /good/ idea :>

 

[Corey Murtagh]

Eh? One gains mass by switching on gravity?

I can almost see it, in a twisted way. With gravity a body will attract
other bodies, which increases the probability that it will collide with
something, and that the collision will produce an amalgamation. Each
such collision increases the mass of the amalgamation of objects.

But damn, no one told me I had a /choice/ in the whole gravity question.
Where's the switch?

 

[Corey Murtagh]

The domestic cat clearly has intelligence from the way
in which it resolves barriers that are placed in the path
of its nighly prowl.
Clearly.

However, that is no evidence of self-awareness. We will
not find the solution to our quests by confusing the issues.

Are you trying to tell me that cats aren't self-aware? My cat takes
exception to that. She's quite aware - if perhaps a little deluded,
from our limited, human point of view - of who and what she is.

You obviously don't know cats.

 

[Roedy Green]

We clearly have something. No *other* animal has taken over the
planet (nor come close to messing up the entire biosphere).

We have hands. This allows us to get into more mischief that most
other species with less agile manipulators.

Destroying the biosphere, surely has to be a sign of pathology, rather
than intelligence. We humans are most proud of that which has done the
greatest damage to our environment. We are a bit like clever children
who have learned to create but not how to control our creations.

Paleontologists measure the success of a species by how long it
survived before going extinct. We humans are a fledgling species
teetering on the brink of extinction already. Most likely we will be a
nanosecond blip in geological time.

One of the most dangerous creations of man was the corporation, a
legal fiction with all the rights of a human, the wealth and power of
many, impervious to most written and unwritten laws of decent conduct,
impervious to punishment, impervious to the death penalty, and no
conscience at all. People feared they would become slaves of their
creations. Wake up gang. It has already happened.

 

[Corey Murtagh]

Read some of the works of the late Stephen J. Gould.

It does not require a god for random selection to have a directed
purpose. Natural selection provides that direction.

Either there is direction, or there is random selection. Which is it?
The two are mutually exclusive, as Richard was trying to explain.

Personally I see no evidence of direction in evolution. There's what
works and what doesn't... and we keep finding ways around what doesn't.

 

[Lee Fesperman]

I think that is true for skilled programmers, and for simple tasks.
But for complex tasks it is more like a jigsaw puzzle, were you put
together bits of the solution by looking at the tools. Gradually as
you create classes, you have better tools to think about solving the
entire problem with. You let your mind stop fussing over details and
think more abstractly with your newly create though tools.

We used to call that bottom-up programming. I prefer a mixture which includes top-down
programming that I call "nibbling at the edges".

 

[Roedy Green]

From what (little) I've seen, they produce very derivative--and
not very interesting--works.

You can see some samples in the Spiritual Machines book I mentioned
earlier. I have NO idea how I would write such a program.

 

[Programmer Dude]

Genes solve the problem of perpetuating their existence, because
genes that don't, don't survive. The genes require a certain
art to pull this off, not non conscious intent. I think you are
equating intelligence with conscious intent + ability to solve a
problem. I am using it to mean merely the ability to solve a
problem.

By this standard, one could say rain water "solves" the problem of
removing the air from a pail.

 

[Roedy Green]

I met a genius, Bernie Till, who built a computer model of a nematode
worm nervous system for his Master's thesis. It had 47 neurons as I
recall. The thing that blew me away was this model demonstrated ALL
the known behaviours of nematode worms.

He vainly tried to explain to me how it all worked with S functions.

There was no other magic that the classic biochemistry of neurons.

Explicit neural nets are an extremely compact way to encode some
pretty complicated algorithms. Imagine how much traditional Java code
it would take to animate a worm in a totally lifelike way, have it
respond to chemicals, seek a mate, navigate obstacles etc.

We know quite a bit about neurotransmitter, how the combine, how
synapses become acclimatised, tired etc.

What we have not explored yet in the mainstream is writing algorithms
with explicit neural nets, to see just what sorts of things they can
do. Someone "wrote" a flip flop using a neural net to store a bit.

The other area coming along in the background are self-programming
neural nets, which adjust their own connections by learning patterns
from examples. We do all kinds of computation in our brains without
having clue how we do it. Neural nets are similar. They work, but no
one need understand how.

 

[Programmer Dude]

We have hands. This allows us to get into more mischief that most
other species with less agile manipulators.

It's more than hands. The apes have hands, and many of their
genetic lines have been around *much* longer.

Destroying the biosphere, surely has to be a sign of pathology,
rather than intelligence.

Perhaps, but notice we've begun to catch on to the danger and
stand of good chance of dodging the bullet.

We humans are most proud of that which has done the greatest
damage to our environment.

Speak for yourself. I'm most proud of art and communication.

We are a bit like clever children who have learned to create
but not how to control our creations.

Yes, we probably are still children in many ways.

Paleontologists measure the success of a species by how long it
survived before going extinct.

Paleontologists probably equate longevity with success. By that
measure, sharks and cockroaches are *highly* successful! I prefer
a more discerning standard.

Most likely we will be a nanosecond blip in geological time.

You appear to have a very cynical outlook! Our history is only
begun; the finish remains to be seen.

 

[Arthur J. O'Dwyer]

We clearly have something. No *other* animal has taken over the
planet (nor come close to messing up the entire biosphere).

There are several orders of magnitude more ants on the planet than
humans, although they haven't "messed up the entire biosphere"
AFAIK - just a few picnics.

Green plants, of course, messed up the biosphere pretty badly when
they evolved. Pumped all that toxic oxygen gas into the atmosphere;
pulled out all the carbon dioxide. Of course, it depends on how
broadly you're going to categorize them as to whether one can say
that "algae have taken over the world," or "deciduous trees have taken
over the world," or whatever.

For real messing-up-biosphereness (following humans), I vote for
rodents. They've managed to get pretty much everywhere, and eat
up lots of indigenous populations.

....the difference is that rats and trees don't care.

-Arthur

 

[Programmer Dude]

The technology is greatly improved now. You might give it another
look.

It's not the technology; it's the *idea*.

I think it's idiotic. Natural human language is very inefficient,
highly context sensitive and incredibly variable between cultures.
It is--flat out--a poor way to communicate effectively.

 

[Programmer Dude]

It does not require a god for random selection to have a directed
purpose.

Don't you think "random selection" and "directed purpose" are
mutually exclusive?

Natural selection provides that direction.

No, it provides a *filter*.

 

[Programmer Dude]

Given that we have no way yet to measure consciousness,...

No? If I say, "Knock, knock," and you answer, "Who's there?"
That would seem to indicate you were conscious.

Eventually a time will come when computers start requesting to be
treated fairly.

There MAY come a time. It's not a given. We have no idea if
sentience is an emergent property of complexity or something
else very special and not mechanically reproducible.

We will be in a quandry to decide if we are being compassionate
or silly to comply.

Or perhaps compassion and silliness will have less to do with it
than rational analysis.

I have watched my own consciousness disappear and reappear with
anaesthesia. So it seems possible that consciousness is just some
quantum phenomenon that happens...

Yes, possible.

...where there is enough activity in a small enough space (a brain).

What would size have to do with it?

I thus see no reason why it should not just manifest to varying
degrees in computer chips.

It *may* require more than sheer complexity. I read a paper once
by a "philosopher/physicist" at MIT who floated the idea that
consciousness might be a "fifth force" (along with gravity, strong,
weak & electromag). Interesting concept, and it still gives me
something to think about.

 

[Eray Ozkural exa]

An upper bound for delta is, for instance, the size of an interpreter
in language B for language A. But delta can be quite a lot smaller
than that.

Well. I think this sentence itself demonstrates how important
pragmatics of a programming language is. If I'm going to write an
ocaml interpreter in the lousy C language, which will be a HUGE code,
and then write my code in ocaml what's the point of using ocaml?

I don't think you can simply analyze this in terms of kolmogorov
complexity. Here, the claim is valid for reasonably written,
comprehensible, conventional programs, not theoretical curiosities
such as minimum program size for a given problem w.r.t. a formal
system.

What the above discussion implies, though, is that for VERY COMPLEX
CODES, you *will* have to develop your own appropriate languages and
down the ladder the most expressive or
human-cognitive-system-compatible language wins. I don't think that's
much of a discovery... So, why am I left puzzled?

I think my theory is that, you are wrong when you say "delta can be
quite a lot smaller than that". That's simply not true when you are
contrasting Java with ocaml *and* that interpreter size will be
enormous. (How so? ocaml is so much more than Java semantics-wise, you
would be stuck in mud trying to realize all of those mechanisms with
Java)

Now, why is this strange? Because I'm getting the feeling that your
claim contradicts some of the "power of expression" results that show
how much better a class-based programming language is compared to a
"flat" language (be it functional or imperative).

What you are missing is that's a minimum condition, and we can't know
in advance if we will be able to compress the code like that. In many
cases, I suspect you will find that the difference will actually be
greater than interpreter size.

Best Regards,

__
Eray Ozkural

 

[Programmer Dude]

What was it that you observed that lead you to that conclusion?

Just decades of being around--and very interested in--animals.

Another way of asking the question is what behaviour do you see in
humans that convinces you they have a sense of past and future?

Cognito ergo sum. *I* remember the past and dream of the future
(and other humans I trust report they do also, but the primary
information is my own experience).

Next I read that I share 98%+ genes with a chimp. We are obviously
more alike than different.

Depends on your definitions. In some regards, we share much. In
others, we share little.

This implies in recreating human intelligence, most of the work
will be in getting us to the squirrel level. Getting PAST human
level will be snap once we have done that.

Yes and no. We still can't replicate the complexity of, say, a
fly (let alone a squirrel). A brain is *awesomely* complex in its
wiring and--more importantly--cross connections.

But *sentience* is a quantum leap that may or may not come from
complexity. If it doesn't, we may never succeed.