AI Mind Philosophy

[Kellogg]

Hey all,

I got a question in my e-mail and thought I'd throw it open for discussion. Be interesting to hear what you all think:

e-mail address withheld for privacy's sake wrote:
> Hmmmm... Interesting notion about AI systems... Intelligent
> performance would be much more effective than a personality
> simulation... How much work have you done on the background
> of all this...? Just wondering, because I'm working on a
> background of my own, and have noticed interesting bits here
> and there in the comic...

Hi!

That's a really neat question.

I'm afraid I really don't have much of any sort of background in
programming, I build computer chips for a living, (or will do
when the economy and job market picks up again) but I've read
AI material.

A lot of attention is paid (publically anyway) to The Turing test.
The Turing test, in my opinion, is just a test of how well a
computer can simulate a human being. Outside the gaming and
entertainment industry, I really can't imagine a use for such a
simulation. I don't think people would want them around for use
in practical situations. Nobody wants a talking car. Talking
cars were experimented with in the 1980s and they quickly found
out that people didn't like 'em! Nobody has a car that verbally
nags them that the lights are on, or the door is open or please
fasten your seatbelt any more. Even if you put an AI in the car
instead of a pre-recorded nagging voice, people wouldn't want it.

If a car should have a computer with the capacity of HAL-9000 in
it, the computer should be concentrating on driving the car, not
wasting computation time and clock cycles on finding out if I want
to play a game of chess.

Now, for some applications that's fine, but I don't think that
should be the primary focus of the development of intelligent
computers.

Think of all the stuff where a computer could really be
used. That's where we need them.

-An air traffic control system that interacts with a plane's
auto-pilot to provide heading and altitude changes that get
the plane to it's destination without having to force the
planes to take long detours to areas where they are more easily
controlled but closer together where they're more likely to
collide. Not to mention taking meteorology into account so that
it could find favorable winds and route you around storms.*

-A computer in a car that not only controls fuel economy and
flow, but can diagnose engine problems and notify you what
maintenance needs to be done and when with more information
than just "Check Engine". Heck, that wouldn't take much in
the way of computer brain power.

-I've worked with some fairly sophisticated computer controlled
manufacturing equipment, and am amazed at just how stupid they
are. The equipment manufacturer provides a troubleshooting
guide in the manual. Why can't the system's computer run through
the logic tree of the manual and guess at the possible causes of
the problem, and provide advice as to what things to check,
instead of something as low tech as a paper manual? (Especially
in a clean room environment where paper can contaminate things!)

-If you want an entertainment computer, how about one that
notes the things I watch on TV, or read on the net, and makes
intelligent recommendations on new stuff, or brings news stories
to my attention that are interesting or affect me.
(Ah, here's an article on materials science he'd like...)
(Oh! He reads science fiction, and comic strips, how about
Freefall, Shlock Mercenary, or 21st Century Fox?) (True, there
are computers that *sorta* do that, but not very well, and *I'm*
not the one in control of it. The ad spammers are.)

Anyway, that's just 3 examples of how a computer could perform
intelligently. They're just based on my experience. I'm sure
you with different experience could come up with other examples.

However, the stuff that gets press in artificial intelligence
is essentually "simulated people". Now, I *hope* that's the
stuff we hear about because it's *neato keen science fiction
stuff* that the press loves to scare the luddite technophobes
with. I *hope* that there's stuff going on on intelligent
computing, but I just haven't seen any.**

I think I'm going to post this to the forum. There are more
than a few intelligent, well read folks out there who may have
their views on the subject.

Hope that answers your question!

Scott

*Note: I really think that this is possible *now*, given the
sophisication of the programs that are out there and what I've
seen of air traffic control. The collision avoidance system
in Europe *knew* that the Russian Airliner and the Cargo Jet
were going to collide, but it still required a person to call
up the jets and say "Hey! Look out!" which they failed to do
in time. We're going to spend Billions of dollars on airport
security at a *single* airport in LA because of one nut with a
gun on the ground who never would have made it into the air.
If a *portion* of that was spent on Air Traffic Control, a
computer system could be put in place that would really save
time, energy, fuel, money and *LIVES*.

**Note 2: I think this is also going on in the cloning debate.
The press likes to scare people with "Grab your torches and
pitchforks! They're gonna clone Hitler!" Whereas, even if the
worst nightmare *did* occur, what would you get? An infant
with a predisposition to get cranky, manipulative, yell at people
and with only one testicle. But, it wouldn't be Hitler, nor
would it mean the beginning of WWIII because it took a lot more
than just one cranky nasty guy to do that. But, the real
applications of cloning would be more like producing replacement
organs that won't be rejected or require the immune system to
be shut off...

[Marc_s]

Outside the gaming and
entertainment industry, I really can't imagine a use for such a
simulation. I don't think people would want them around for use
in practical situations.

In Computer science, as with all science, the decisions are made by
corporate management. ( With valuable data from sales and marketing. )
To them, AI was going to eliminate all kinds of jobs, which would mean
bigger executive salaries. Expert systems were going to replace specialists,
and intelligent agents ( IA ) were going to be virtual personalities.

When management realized that science was overly optimistic, they stopped
funding AI development. The corporate world as a whole has stopped research
in general. There is more short term return in putting that money into marketing.

That is not to say that AI has had no effect. Most of you have disabled a widely
recognized intelligent agent. That annoying talking paperclip in 'Word', brings
new meaning to the term "killer app".

We haven't seen true fuzzy logic, but it is fuzzier. A lot of banking and insurance
clerk jobs have been reduced. Some tech help operators have been replaced
with agent type systems, as have telemarketing operators. But more telemarketing jobs
have gone to cheap prison labor, than have been replaced with AI.
In the case of the technical jobs, it was cheaper to lobby govt. for more
cheap foreign labor.

Where AI and fuzzys have really made a difference is things like voice recognition,
and sorting through large amounts of data. ECHELON, ( monitors satellite, microwave,
cellular and telephone conversations ) and Carnivore, ( reading this, and all your e-mail now )
are the cutting edge in AI right now. At least Big Brother still funds R&D.

So, not to worry. We have nothing to fear from "simulated people".
I'm not even worried about Big Brother. But there is no greater threat
to freedom than telemarketing, and that evil paperclip.

[ZOMBIE USER 6611]

In Computer science, as with all science, the decisions are made by corporate management.

I will take the liberty to post an except from a bit of fiction that Mr. Kellogg is helping me with. This chapter takes place in 2019.
----------------------------------------------------------------------------
Chapter 28: Intelligence Design
Michael enjoyed musing on the topic of artificial intelligence. He had absorbed, with great interest, many books on the topic, from Asimov's "Robot" series to the speculations by Roger Penrose at the turn of the century to much more recent work.

In Michael's estimation, the recent work tended toward two extremes: the purely mechanical implementation strategies, or the purely philosophical expositions - rants, even - by those not qualified to understand the engineering aspects. To Michael, both of these were unsatisfying.

Over the years, the power and speed of computing systems continued to advance. The size continued to reduce. And as these systems became more and more powerful, they finally reached the attention of groups of persons who felt that they should not be allowed to progress further.

These were dismissed as crackpots early on, although since 2008 the prohibition of computer intelligences had been a plank in the presidential election campaign. The concerns were certainly premature in 2008, but they had their impact nonetheless.

Humans seemed predisposed toward fright. They wanted to believe something bad. Genetically enhanced food production had been a necessary part of the world's production since long before the turn of the century, but there were still folks who believed that it was the food planted by Satan himself.

Global warming was handled in a similar fashion; by distorting data and emphasizing only certain points, the media portrayed to the citizens of the world a grim picture indeed. Many still clung to the "hope" that global warming would threaten the human race - but the reality was that the changes were mostly beneficial. The powers that be realized this when it became politically correct to note that the increased CO2 in the atmosphere was preventing millions of people from starving; this "poisonous greenhouse gas" was in fact the most vital part of plant growth. It was known thirty years ago, back in 1990, that crops were more productive as a result. The dreaded "killer storms" were a disappointment for many; they didn't happen.

Still, America had been ultimately shoehorned into what was now "Kyoto III", although it was not the Kyoto version of this infamous treaty. Developed nations produced greenhouse gasses, so the logic went. Developing nations did not (ignoring the data). So, the logical solution was to pump billions of dollars and euros, and trillions of yen into the economies of the developing nations so that they could be "polluters" too.

Michael had no official opinion on this. He noted, with well honed realism, the political results and the costs. But, the developed countries signed up for this, so in theory it was deserved.

An offshoot of the Greenhearts felt that it was their moral obligation to oppose further work on the development of computer intelligence. At first, comics and editorials ridiculed their efforts. As they became more "politically correct", as the old phrase went, the ridicule turned onto first the computers, then those that worked in their design. It was a profession now the target of much humor, as attorneys had been for hundreds of years before. Many brilliant people decided to pursue scientific careers that were not so much against the tide of the populace.

At the end of the last century Roger Penrose, in "The Emperor's New Mind", described in great detail speculations and theories concerning artificial intelligence. He proposed that the neurons that occupied brain cells contained some properties that had not yet been discovered by physicists, and may never be discovered.
And that "thinking", or "self awareness", depended upon this nearly-mystical set of properties, and therefore was not going to be reproduced in an artificial construct.

Other workers who actually designed such things for a living often had a different opinion. While not approaching the subject with Penrose's trademark brilliance and thoroughness, they felt that self-awareness was something that would naturally develop as the system reached a certain level of complexity. The early work with neural networks was promising, but Penrose remained convinced that nothing fashioned by the hand of Man would ever possess self-awareness.

Penrose, thought Michael KOH-AI v3.02, was wrong.
----------------------------------------------------------------------------

[Icefox]

More uses for AI:

Visual recognition software. That could be used for airplane or car navigation, or maybe law enforcement, or aids for blind people, or smarter/better robots. This usually falls under AI 'cause the computer has to make assumptions about what it's looking at, and learn about new things.

This was sort of inspired by you talking about air-traffic control, but what about ground-traffic control? Aren't you sick of traffic jams?

LevelHead, that sounds like a really cool story you're writing. ...I don't suppose you need another editor? *beg, plead* Heh.

...I suppose that one of the biggest problems is, as LevelHead mentioned, fear. Having a unique device of immense power being able to make it's own decisions is freaky. What if it makes the wrong decision? People never stop to think that the heads of governments and companies make their own decisions too, independant of anyone else, and that their decisions are often wrong also. But the fact that it's not human makes it even worse. I mean, look at how scared humans have been of other humans, just because they were different.

...Was that at all coherant? Lemme simplify it...

BAMPH!
*turns in to Yoda*

Misunderstanding leads to fear. Fear leads to anger, anger leads to hate; that way the Dark Side lies.

[Kellogg]

The Evil Paperclip!

Yes, Lindy has his origins based more on that paperclip than Hal-9000.

Level Head's story!
Great stuff! (Though, I confess to being behind on reading the new stuff. I'm behind on Wednesday's strip too!

The Emperor's new mind sounds like he's trying to include a soul at the level of molecular biology. What was it they're called "Medichlorines"?

As for fear of autonomous creations:
Go read Mark Stanley's Freefall for the last few strips! For the last few strips he's been discussing how a genetically engineered wolf (Florence Ambrose, mentioned in Wednesday's strip) has "Safeguards" built into her to make certain that she's "Human Safe".

http://www.purrsia.com/freefall/

Okay, now I really gotta go work on Wednesday's strip!

Scott

[Kellogg]

An offshoot of the Greenhearts felt that it was their moral obligation to oppose further work on the development of computer intelligence. At first, comics and editorials ridiculed their efforts. As they became more "politically correct", as the old phrase went, the ridicule turned onto first the computers, then those that worked in their design. It was a profession now the target of much humor, as attorneys had been for hundreds of years before. Many brilliant people decided to pursue scientific careers that were not so much against the tide of the populace.

This actually reminds me of a conversation I once had with some Physics Grad Students from the Soviet Union. (I say Soviet Union because this was a couple months before the breakup.)

We were talking about computers and computer games and they were telling me that their favorite computer game was F-19, where you flew an American stealth fighter in to attack the Soviet Union. This raised my eyebrow quite a bit. But, it became apparent they were not very happy with their government (to say the least).

Anyway, beyond this, one thing that had always puzzled me, so I asked: Why is the Soviet Union so far behind the west in computer technology? They've got smart people who worked hard, so why were they copying obsolete Western technology instead of developing their own?

Well, the answer came back that the Soviet Union was *way* behind in computers. The why of it was because many years ago, Stalin had declared computers to be a "Bourgeois" science, and anyone who studied it would likely be picked up by the KGB.

As Level Head put it, computers were just not "Politically Correct". Literally.

Scott

[Quill]

One use for an intelligent system could be to run a house or other residence. I would love to have something like that...one central computer to control a/c, entertainment, phone, and possibly the bath and the microwave. It could let the dog in and out on demand, feed the cat if one was getting home late, and make sure the lights come on when it gets dark. Not exactly an important use for such a system, but I think it would be fun.

This could also alert medical or other authorities if something went wrong with the inhabitant(s).

[ZOMBIE USER 10683]

First, I want to say 'Hi!' to Scott and the rest of you. There, now that I've gotten that out of my system:

A common problem designing simulations is that people, even experts, don't really know what they are doing. As a result, they are at an immediate disadvantage when they try to teach a machine to do something. By this I mean that the way the stated task is explained to the machine (something like: maneuver my tank battalion to defeat the enemy, for example) doesn't bear much relationship to the way a person would actually accomplish the same task. So there is disapointment with the AI result. One benefit of trying (even though we keep failing) to design an AI os better understanding of the real task.

Anyway, my $0.02.

-otrstf (Dan)

[Baxtrr]

One use for an intelligent system could be to run a house or other residence. I would love to have something like that...one central computer to control a/c, entertainment, phone, and possibly the bath and the microwave. It could let the dog in and out on demand, feed the cat if one was getting home late, and make sure the lights come on when it gets dark. Not exactly an important use for such a system, but I think it would be fun.

This could also alert medical or other authorities if something went wrong with the inhabitant(s).

"The front door started to scream at me. "Dirty feet, muddy feet! Wipe your feet! Please be NEAT!" I shot the damn thing in its keyhole."

--Ray Bradbury

bax

[Kellogg]

First, I want to say 'Hi!' to Scott and the rest of you.

Welcome aboard Otrstf!
Please insert: Shameless Plug for Otrstf's story, Anthropomorphic Foxes In Space
http://home.earthlink.net/~otrstf/

In brief: A UFO/Roswell alien story done *right* for a change.

I'm not usually a fan of the genre, but Otrstf's got it right.

A common problem designing simulations is that people, even experts, don't really know what they are doing. As a result, they are at an immediate disadvantage when they try to teach a machine to do something. By this I mean that the way the stated task is explained to the machine (something like: maneuver my tank battalion to defeat the enemy, for example) doesn't bear much relationship to the way a person would actually accomplish the same task. So there is disapointment with the AI result. One benefit of trying (even though we keep failing) to design an AI os better understanding of the real task.

Very true, but I think that the situations where things that require creative decision making are less common than is obvious.

Setting up a battle plan for a tank battalion requires creativity in that you don't want everything to be predictable.

On the other hand, a troubleshooting guide for mechanical or electronic problems often have logic flow charts. Aircraft pilot's manuals have checklists.

There are a lot of tasks for which an AI would be very suited, others not so. But beyond computer games, newscasters, pop-stars, and politicians, I can't think of a decent application for a "Simulated Person" (And I'm not sure if 3 out of 4 of those applications are "decent" either.

[Kellogg]

"The front door started to scream at me. "Dirty feet, muddy feet! Wipe your feet! Please be NEAT!" I shot the damn thing in its keyhole."

--Ray Bradbury

Ancient Anecdote Time!

Back in the 1980's, Marc_S's father use to own one of the first Talking Cars.

It didn't say much, except to tell you that you'd left the headlights on.
I'm not sure, but I think I was the first person to notice that it was a female voice saying:
"Please turn out the lights..."
"Please turn out the lights..."
"Please turn out the lights..."

Or maybe it was just the teenage hormone factor kicking in.

Anyway, Marc, who is very good at vocal impressions, and noises, secretly recorded a cassette tape, and stuck it in the car's tape deck so that it would turn on when the ignition was turned over.

"I didn't like the way you drove me last night..."

Scott
(Anyone remember the old Twilight Zone episode "Living Doll"? "My name is Talky Tina, and I'm going to kill you..."!)

[ZOMBIE USER 8834]

I, too, had a talking car through High School. It was my second car, and rather than giving it a name, if I wanted to refer to it personally, I used a nasty epithet. "Chrysler" also became a swear word to me. It was an '87 LeBaron 4-door hard-top.

And the voice was male-- and crabby.

My best friend and I tried one day to dig it out, and went all the way through the glove box into the firewall and could not find the source of the voice. Much to our chagrin.

I still shudder when I think about that thing.

Ulgh!

--TK

[Kellogg]

And the voice was male-- and crabby.

Eww... What did it say?

Scott

[Marc_s]

And the voice was male-- and crabby.

Eww... What did it say?

Scott

"I don't like pits! pits! pits! in my juice! juice! juice!

[ZOMBIE USER 6611]

And the voice was male-- and crabby.

A true story about an irritating computer voice. Friend Talon, you know all of this already, but our other friends here do not.

Last November at the Playboy Mansion a big celebration was held in honor of Arthur C. Clarke, the writer of book and movie "2001". Sir Arthur attended by full-body two way hologram from Sri Lanka. Patrick Stewart was the major host, but the other host was HAL 9000 from the movie -- and on stage was the control panel with HAL's very recognizable red-eyed port scanning the audience.

HAL's job, apparently, was to keep the large number of celebrity guests to their allotted time. Several of these celebrities were visibly annoyed at HAL's interrupting their speechs and shooing them off the podium for the next guest. At one point Buzz Aldrin looked angry enough to look for a plug to unplug, and Patrick Stewart's patience was clearly being tried.

It was supposed to be an act, of course -- but the HAL actor took his role rather seriously, and even the audience, after laughing at the first few interruptions, later were cringing at Hal's "Excuse me, Patrick...".

It would not surprise me in the least to learn that angry words were exchanged backstage; I did not inquire.

It didn't matter who the celebrity was, HAL would step on him or her like a trucker with a thousand watt CB. Even James Cameron, director of the Terminator movies, Abyss and others, whose tribute to Sir Arthur had Sir Arthur visibly choked up.

But HAL was human after all; Morgan Freeman took over the podium, reading from Rendezvous with Rama and explaining his next project which was to make this movie. HAL was too absorbed, apparently, to interrupt. The audience didn't mind a bit.

[Baxtrr]

And the voice was male-- and crabby.

Eww... What did it say?

Scott

"I don't like pits! pits! pits! in my juice! juice! juice!

Yeaggh! ODD COUPLE Flashback! Yeaggh!

bax

[Kellogg]

Eww... What did it say?

"I don't like pits! pits! pits! in my juice! juice! juice!

Yeaggh! ODD COUPLE Flashback! Yeaggh!

Good GRIEF! *THAT'S* where it came from!

That' was driving me *nuts*! I could hear the voice in my head, but
I didn't know where it came from!

Scott

[Kellogg]

Last November at the Playboy Mansion a big celebration was held in honor of Arthur C. Clarke, the writer of book and movie "2001". Sir Arthur attended by full-body two way hologram from Sri Lanka.

Good old Arthur C. Clarke.

I think it was a few years back, I read an interview where he suggested that a car made with buckytubes for structural materials would weigh in at 1/100th of the current weight of cars.

Meaning a 4000 pound car would weigh 40 pounds. Being the conservative engineer that I am, I decided that 500 pounds would be
enough of a weight saving. That's why the T-Bird is as light (or heavy
depening on your point of view) as it is.

Scott