Re: How Robots Will Steal Your Job

R

Roedy Green

Were you not expecting emergent behaviour from a complex system?
Not back then. My image of the computer was something I completely
controlled, something that allowed me to push for and sometimes attain
perfection because I could have as many tries as I wanted, and it made
no record of my previous attempts. It was not like a mechanical
drawing that deteriorated with each erasure and change.

I did not think of my program as being any sort of AI. I knew every
line it in thoroughly.

I had been surprised by programs before in that they may turn out to
be more useful than I expected, that users would find novel ways to
employ them. I had seen erratic and interesting behaviours in buggy
programs, but this was something coherent that LOOKED as if somebody
had done a great deal of work planning it. It was as if somebody had
been monkeying with my program in my sleep to add higher order
features to it.

People often baldly assert there are no such things as emergent
properties, e.g. that a computer could never have anything like a
personality or an artistic style so I don't think many programmers
have yet had this experience.
 
R

Roedy Green

Although, sometimes, op-
timum solutions are both possible and essential, as in the
Apollo Lunar Module's guidance and control system.

We have something vaguely in common. JPL used my 32-bit BBL Forth
interpreter to write the code for some unmanned missions. How did you
get such an interesting job? One thing many people don't realize is
that space missions use hardware of considerably older design than the
rest of us.
 
G

George W. Cherry

Roedy Green said:
We have something vaguely in common. JPL used my 32-bit BBL Forth
interpreter to write the code for some unmanned missions. How did you
get such an interesting job?

I applied for it! The MIT Instrumentation Laboratory (now
the Draper Laboratory) was the prime contractor for the
Apollo Navigation, Guidance, and Control System. I joined
the Laboratory in 1961 to work on the ANGCS and be-
came the project manager of the Lunar Module NGCS.
The computer was a very interesting one, but the algorithms
and programs were what made the project a success. I
designed the Lunar Module digital autopilot. It was the first
digital autopilot ever. It was far superior to Grumman Air-
craft and Engineering's back-up analog autopilot.

Apollo was a moondoggle. It is a mark of Bush league's
strange "intelligence" that he wants to "establish a presence
on the moon". The purpose of Apollo was to one-up the
Soviet Union in space. I didn't care about that, though: I
just wanted to do aerospace engineering on a challenging
project that was not military.

George
One thing many people don't realize is
that space missions use hardware of considerably older design than the
rest of us.

The Apollo spacecraft computer was really quite innovative.
 
O

OmegaZero2003

Wonderful Java Glossary there Roedy!! I am surprised I have not stumbled
upon it before. I have been working with Java/J2EE/JavaSpaces/etc. for
several years. I based an Intelligent Agent System on that platform (along
with KQML).

AFA emergent properties are concerned I agree with your take; synergy has
many exemplars in nature and reductionism is a lossy process in many cases.

But that you found programs more useful than thought/intended is a product
of both the program + user = system, not the program itself, right!
Implicate features are not necessarily features those that are characterized
as emergent when explicated. Rather, emergent features are not only
surprising, but are unpredictable from a reductive analysis of the parts and
their relationships.
 
R

Roedy Green

But that you found programs more useful than thought/intended is a product
of both the program + user = system

What I did was invent a number of generic utilities that worked on
files. What my users did was create giant incredibly complex jobs
weaving the utilities together in an elaborate web. They treated each
step with the same casual abandon I would add two variables in a
program. Each utility became a programming atom to them. When I
created the suite I never envisioned more perhaps half a dozen being
strung together. Part of the problem was my eyes were focussed too
much on the low level. I had a failure of nerve to use the suite in
such a high level way. I would likely have resorted to custom
programming to solve their problems, rather than using a somewhat more
indirect approach using the standard utilities.

We were working on a large lawsuit. BC Hydro had been sued by the
contractors on the Peace River Power Project. We had rooms full of
punch cards (moved to tape). The lawyers would demand some
statistics, graphs or information, and we had to come up with in it
with a couple of hours notice. This was in the days when you
typically got one batch run per day on the mainframe. So the utility
technique worked very well since the jobs usually worked first time.
The goal was to shorten coding time, not execution time.
 
Z

Zagan

Sir Charles W. Shults III said:
Making such a program is nearly trivial. Conway's Game of Life is proven to
be unpredictable in the sense that you can only see what the outcome will be by
running it. In the same vein, you can easily make a program that shares this
property. Therefore, making unpredictable results in self-modifying code can be
almost guaranteed, given a little unsupervised running time.
As for the question, "who created the final outcome", that is for the
philosophers. It is true that you made the efforts that produced the program
that wrote the program, and if it is done on a computer using pseudo-random
number generators, then it was wholly and completely deterministic- so, you are
directly responsible. But that does not mean that you envisioned the end
product!
Sort of like dropping a jar of jam from 10 stories up and knowing what the
splat would look like.

Cheers!

Chip Shults
My robotics, space and CGI web page - http://home.cfl.rr.com/aichip

[Zagan]
If I may insert my 2 cents here... :*)

As Chip noted, " 'who created the final outcome' is for the philosophers." I
tend to agree with this, but I will put a differenct twist on it below.

But before I do that, I want to make the point that it seems most people
confuse intelligence and consciousness. While there is certainly a
connection between intelligence and consciousness (as demonstrated by those
entities that possess both), I must insist that intelligence can exist
without consciousness. Is a paramecium conscious? It, being a single cell
organism, I doubt it. But yet it does display intelligence (however
primitive that may be). Consider a colony of bees. Is the qween or worker
bees conscious? Again, I doubt it, but yet the colony does exhibit
intelligent behaviour and produce complex structures.

The world is full of life forms that exhibit intelligence, but not
necessarily consciousness.

If we take this as a given, then we might ask the question, "where does this
intelligence come from?" The religious-minded may say, "God did it." The
non-religious may say the intelligence is a product of evolution. If a
religious scientist and a non-religious scientist were to study a colony of
bees or ants independently, they will likely reach similar conclusions in
their research. The point being that the "source" of the intelligence is not
relevant. Bees exhibit intelligent behavior no matter whether "God did it"
or evolution.

Returning to the issue of programming, we can say that the intelligence
exhibited by a program is that of the programmer(s) who wrote the program.
And this would be true in a sense. I recently worked on code for a rather
large medical program that responded to external hardware as well as user
input. It could be said that the intelligence of the program is my
intelligence. However, I no longer work for that company. Eventually other
programmers will modify the code I wrote; adding features and fixing bugs.
Thus additional intelligence will be added to the intelligence I put into
the code to begin with. The program runs and works according to the intended
design. I don't have to be there for this to happen. This last statement may
seem pointless, but think about it.

The program works as intended regardless of who wrote the code. It could be
said that the intelligence in the program is a result of the fact God
created me, or that I am a product of evolution. My point is that it doesn't
matter!

I'm not saying that the source of "intelligence" is not of interest, I'm
simply saying it doesn't matter in our discussion of "what" is intelligence.

Comments welcomed.

// Jim
 
P

Programmer Dude

Gerry said:
Can anyone obey the three laws of robotics? It appears to me that,
even if we could agree on the definitions of "harm" and "human"
(and we can't) situations would arise in which any action or
inaction would change which individuals were harmed.

In fact, in the later books (not written by Asimov), the robots
design a "Zeroth Law" above the other three, and the Zeroeth law
places *humanity* as paramount. Thus robots can harm individuals
if humanity is served.

Of course, the fundimental moral questions remain: who decides
what's Good and what's Harmful. Oh, and what is "Humanity"....
 
A

Alf P. Steinbach

In fact, in the later books (not written by Asimov), the robots
design a "Zeroth Law" above the other three,

Whoa! Stop right there! _Asimov_ invented the "Zeroth Law" (or
rather, let Giskard the robot invent it).

Any aspiring programmer should know that.
 
P

Programmer Dude

Alf P. Steinbach said:
Whoa! Stop right there! _Asimov_ invented the "Zeroth Law" (or
rather, let Giskard the robot invent it).

Oops! So it was. I got confused by all the Foundation and Robots
books done after the "originals".
Any aspiring programmer should know that.

I'm more perspiring than aspiring at this point in my career... (-:
 
R

Roedy Green

Thus robots can harm individuals
if humanity is served.

It would seem to me any non-human intelligent species would not have a
fanatical loyalty to humans. Their loyalties would be broader to
include more species, perhaps all species, or perhaps only their own
species.

At this point in my life, I think it most probably that man will
destroy himself within a century, and possibly take most of life on
earth along with him. There are just so many avenues to destruction
now open.

An intelligent species, or intelligent creation, even one that had
man's best interest at heart, may be forced to poke a stick in his
spokes to prevent him from destroying the whole planet.

So our intelligent creations might, with the best of motives,
deliberately screw up our technology in order to slow us down.

Humans are too stupid to live. They decide issues by loyalties and
emotions. That is fine when all they wield is a spear, but it simply
won't work when they have bioterror and nukes, and the power to alter
the entire biosphere at the stroke of a pen.
 
C

CBFalconer

Programmer said:
In fact, in the later books (not written by Asimov), the robots
design a "Zeroth Law" above the other three, and the Zeroeth law
places *humanity* as paramount. Thus robots can harm individuals
if humanity is served.

This was the direct result of switching the Positronic software
code from Pascal to C in one of the upheavals at US Robotics.
Rather than recode all their indexed accesses to satisfy the
rigidity of C index basing, they simply introduced a further law.

Luckily by the time they did that most of the problems had been
worked out, so the Pascal to C translation introduced very few new
bugs. The translation also avoided the wild and crazy guys
indiscriminately wielding sharp pointers in illegal manners.
Since further development then stopped the lack of sub-ranges and
range checking did not affect any further code changes, because
there were none.

I got all this directly from R. Daneel Ovilaw as told to Susan
Calvin.
 
M

Max MASTAIL

George said:
Right, but intelligence would be a measure from
results of tests iterated over all such "comparisons".



I think we can resume the concept ant generalize
and define the intelligence as the capacity to store
mental (Virtual ) images * eventually and the faculty(power)
to treat(manipulate) them for react to environnement stimuli

* Virtuality representation of Reality eventually.
 
R

Roedy Green

All of them? Even you?? (-:

Yes, as a species. Every day I try to figure out what I could do or
say that would save my species, but so far I am not having much
effect. I am not smart enough.

I have had a fantasy ever since I was quite young that someday I would
have an artificially intelligent companion that would coach me on what
to do and say so that I could.
 
T

Thomas Schodt

I have had a fantasy ever since I was quite young that someday I would
have an artificially intelligent companion that would ...

"prime" (Sci-Fi, Peter F. Hamilton, Fallen Dragon).
 
R

R. Steve Walz

-----------------
Emotions are the part of thought that escapes our own self-analysis,
but they are, nonetheless, logically derived, we simply have no direct
overview of precisely how we arrive at them, not enough extra processing
power to both reason our way to them, and also analyze that reasoning
itself.

Such a limit is true of anything finite.

--------------
Animals are NOT conscious, because if they were, and I mean in the same
manner as we call it so, they would actively seek our recognition and
company and try desperately to win our acknowledgement of their
awareness so that we would not eat them or let them die.

Do they have something below the sophistication of what we call
consciousness? Surely, and many levels of it besides, but if you can
attempt to imagine yourself failing to realize the importance of
wanting humans to know you are a thinking being, then THAT IS the
level at which they are NOT working, and that seems quite a ways
down the ladder from us, too far, in fact, to recognize as anything
LIKE ourselves anything THAT unaware.

They would be active, but NON-CONSCIOUS, and so they are, except
perhaps for mere glimmerings in apes, elephants, and cetaceans.

I would define intelligence as the ability to solve new problems. By
this definition, computers are not intelligent because they cannot
program themselves.
--------------
Humans can't either. They are "programmed" by their experiences.

What you're fishing for is consciousness, which is the internal
self-modeling in an alternate real space produced by interrelations
of non-contemporaneous memory accesses.

Isn't intelligence something gradual? I mean, since humans (by
definition intelligent) evolved from amoebae (not intelligent), at some
point along the way, a threshold of "intelligence" was crossed. But I
don't think it would be like flicking a switch. Rather, I think
consciousness, like intelligence, is on a greyscale.
------------------------
Nonsense. It achieves subtlety the same way the brain has done, by the
evolution of new kinds of brain matter, so that we need new kinds and
uses of memory to achieve any single step-wise leap of the break
barriers of potential to install yet a higher level of subtlety.

Consciousness is a matter of internal modeling of oneself in a new
kind of space, an imagination, which requires a separate attention to
the very self that is perceiving, as yet another object to consider.

Animals don't realize what we are doing, because nobody can tell them,
and they are probably insufficiently intelligent.
-----------------
They don't realize what THEY are doing, let alone what WE are doing.

And given that
"intelligent" humans can't be bothered to rectify the planet and create
a just society, why should animals bother? Animals just look out for
themselves, just like humans. No, human intelligence still has a way to
go, at least judging by the chimp the Americans chose as their president.
----------------------
We have improved things somewhat over the random luck of the individual
in this new level of subtlety by acting as a group, which can choose to
alter outcomes, while the individual is determined and programmed by
its experiences, and is unable. You see, we can change others, and act
as a group to change others, but we cannot act to change ourself
simply because we ARE ourself, and that's circular and futile, we
are already what we are and cannot change lest we have been changed
from without. There is no such thing as "Free Will". It's nonsense.
Whatever we think we cannot change by whim, or else we would have
already.

-Steve
 
R

R. Steve Walz

Bent said:
Well, they can, but someone has to make the program that programs
them.

I am sure you can make a self-modifying program that is sufficiently
complex that you cannot predict how it is going to end up. In this
case, will it be correct to say that it was _you_ who created the end
product?

Cheers
Bent Dalager - (e-mail address removed) - http://www.pvv.org/~bcd
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

No members online now.

Forum statistics

Threads
474,444
Messages
2,571,709
Members
48,796
Latest member
Greg L.
Top