science

November 22, 2004 9:12 AM

Computer as author. (NYT)"Dave Striver loved the university - its ivy-covered clocktowers, its ancient and sturdy brick, and its sun-splashed verdant greens and eager youth. The university, contrary to popular opinion, is far from free of the stark unforgiving trials of the business world: academia has its own tests, and some are as merciless as any in the marketplace. A prime example is the dissertation defense: to earn the Ph.D., to become a doctor, one must pass an oral examination on one's dissertation. This was a test Professor Edward Hart enjoyed giving." by Brutus.1
posted by semmi (16 comments total)

Is is just me, or does anyone forsee another Racter controversy?
posted by Smart Dalek at 9:43 AM on November 22, 2004

This is kind of an old story. isn't it?
posted by Gankmore at 9:57 AM on November 22, 2004

I was one of Bringsjord's students at the Minds & Machines lab from '98-'00 - Brutus sounds like it's come an awfully long way in the intervening years. The stuff it used to churn out was pointless, barely-English manglings of text.

I still disagree with Bringsjord's fundamental approach, of course - even if I can't blame him for it owing to his being a logician first and foremost - humans simply do not think via logical propositions and there is no working model for intelligence thusly based. Our only working model of intelligence is neural in nature - and thus the only AI approach with guaranteed eventual success in this kind of endeavour is one that attempts to artificially model that which we already know works - the human brain.
posted by Ryvar at 10:38 AM on November 22, 2004

The human brain works?
posted by TwelveTwo at 12:00 PM on November 22, 2004

Ryvar: I still disagree with Bringsjord's fundamental approach, of course - even if I can't blame him for it owing to his being a logician first and foremost - humans simply do not think via logical propositions and there is no working model for intelligence thusly based.

I disagree. The world and the brains inhabiting it are purely logical, meaning that the universe's structure is embodiment of Logic. Of course, we may not know all the intricacies of logic, but that's a practical argument, not as essential one.

Morever, the problem with current logic-based AI approaches is probably with the scope of their domain, rather than the operational algorithms, meaning that all phenomena/qualia (sight, sound, emotions...etc) should be input factors and not just the threads of rhetoric ('rational discourse'). Illustration: If a person has to choose between actions A and B, and chooses A despite intellectually knowing B is 'better', that would be considered, strictly speaking, an illogical choice. But it wouldn't be, because although emotions are unconsidered to be unlike 'rational' cognition, the same corticial algorithm generates them as does sensory processing or motor controlling. In that sense, if & when a human brain is successfully modelled, it will be seen that the base operators are logical in nature.
posted by Gyan at 12:27 PM on November 22, 2004

I think the real key to human writing is the novel metaphor. For example, one of my favorite opening lines:

The sky was the color of a television, tuned to a blank channel. (I probably got part of it wrong, but you get the gist.)

This is a great opening line because it makes a relationship to the natural seen as technological. It foreshadows the protagonist's mood, and his temporary disability that drives him to take the risks asked of him.

The quoted computer-generated paragraphs just don't seem very impressive to me.

"...its ivy-covered clocktowers, its ancient and sturdy brick, and its sun-splashed verdant greens and eager youth"

Is there anything in here that is not a cliche?

Among other problems, the paragraph falls short of the "show me, don't tell me" maxim:

he university, contrary to popular opinion, is far from free of the stark unforgiving trials of the business world: academia has its own tests, and some are as merciless as any in the marketplace. A prime example is the dissertation defense: to earn the Ph.D., to become a doctor, one must pass an oral examination on one's dissertation. This was a test Professor Edward Hart enjoyed giving."

The light shone green through the wavy lead-bound panes of the ivy-bordered window at the end of the hallway. Dave Striver shuffled nerviously in the doorway to the Franklin Metz Reading Room where the department faculty gathered to hear the presentation that would mark the beginning, or end of his carrer. Through the door the muffled voices of his commitee were barely audible as they discussed his oral dissertation defense. He stared at the layered paint on the plastered walls, different shades of white edging a dusty fire alarm pull that had been retofitted into the building, he reached out and caressed this tangible signal of constant use over the passage of time. From inside he heard the voice of Professor Edward Hart, "Ok, let's get this started." The sound of Dr. Hart's voice conveying an eager joy at this final confrontation with a student.

That was my attempt at a rewrite, and I'm just a slightly better hack than the program.
posted by KirkJobSluder at 12:44 PM on November 22, 2004

Yuck, bad grammar out the wazooo. John Waters had an interesting comment about creativity, "never show your first draft to anyone, show them a much later draft and call it your first draft."

Of course, John Waters also said. "Never sleep with cast, crew is better."
posted by KirkJobSluder at 12:52 PM on November 22, 2004

The story, tech-wise, may be old, but in the meantime Brutus.1 learned a few things, and besides, writers (at least non-sci fi writers) are only beginning to consider the implications with regard to human exceptionalism and the nature of creative expression.

I am keeping track of how many articles on the subject include the Requisite Steven Pinker Quote and Requisite Reference to Typing Monkeys.

The bigger question, however, is this: will there be a Brutus.1/Jonathan Franzen literary smackdown a la Deep Blue/Kasparov? If so I hope it happens at Caesar's Palace and that Don King is involved.
posted by foxy_hedgehog at 12:55 PM on November 22, 2004

Gyan, not to derail from the article too much, although I think the issue of AI is very much in play, but you seem to have several definitions of 'logic' in your statements with some problems attached to each...

Logic is first given as equal to a brain and the structure of the universe. I think it's fair to say that the structure of the universe could either be considered space and time (in the abstract) or the totality of infinity (in the specific). For logic to be either of these is an unfalsifiable hypothesis. What I'm getting at here is that to define either logic or the universe to be equivalent is to remove most any meaningful content from a term like logic. It's akin to saying 'Logic is everything' or 'Logic is not only infinite, but all things within infinity.' For me, logic would then no longer be a useful term, as there is not a lot I can say or decide about infinity.

Then, logic is given something like its standard definition (the example of the illogical choice) : i.e. a set of rules for thinking according to reason (this is, of course, a tautology). So, logic is a set of rules, which produce certain results might be better here. I would argue logic is the set of rules that, when applied, produce repeatable, verifiable results (hence its use as the structure for the scientific method and also mathematics being a straightforward application of logic). As you point out, then, an emotional choice can easily be an 'illogical' choice. Perhaps 'belief' will work better here than choice, as a choice cannot in itself be illogical, but rather the beliefs behind it can. So, someone can believe that one thing can be something it is and something it is not at the same time (Logical Principle of Non-Contradiction). This belief would be 'illogical.'

Then you seem to define logic as those thoughts (or just phenomenon) produced by the human brain. An emotional choice (belief), you say, is not illogical, as the brain is the thing producing the choice. Here, you describe all thoughts as stemming from algorithms operating in the brain. Again, I would argue the word 'logic' would become less useful defined this way - If all phenomenon in my head are 'logical,' then what could I conceive of (and hence, know or experience) that is 'illogical.' Here, I should state that terms like 'infinity' and 'thought' are relatively synonymous for me (thought being all that happens in my head - as 'all that happens in my head' is 'the sum of all that I can ever know or experience').

Some interesting ideas, not trying to pick a fight... Most curious to me is your statement that brain algorithms produce all thoughts. While brain scans and other imaging and observational strategies have produced a correlation between certain admitted mental states and certain brain activities, all that has been shown is, well, a correlation. Further, the kinds of mental states one has during these observations is rather broad and based on the ability of an observee, or test subject, to communicate 'what they were thinking.'
posted by Slothrop at 6:32 PM on November 22, 2004

"They are perfectly capable of nonfiction prose, and while the reputation of Henry James is not yet threatened, computers can even generate brief outbursts of fiction that are probably superior to what many humans could turn out - even those not in master of fine arts programs."

SUPER SLAM!
posted by Optimus Chyme at 6:57 PM on November 22, 2004

Slothrop, I may have been unclear in my comments, as they normally are, when written on spur of the moment.

Let me rephrase my beliefs on the matter.

1)The universe is structured. Call this structure S.
--
2)We are products of the universe. In other words, everything you are and everything you perceive and interact with, are manifestations of S.
--
3)Certain human activity aims to elucidate S by embodying it in a structured framework of symbols(call it L). This process is continuous and evolving. As far as this process is repeatable, consistent and invariant, we ought to believe that it is approximately accurate and essentially clear. This relies on the infallibility of induction, but we have no other choice in the matter.
--
4)We hold certain qualia('rational discourse') to be representative and isomorphic to L, while others(emotions) not to be. That emotions have their own "logic" behind their operation.
--
5)You talked about mind< ->brain just being a correlation. That's true of all science. If A happens and B always follows and if B never happens if not preceded by A, we say A is the sole cause of B. We might provide some narrative of intermediately-correlated events and label it a 'mechanism' , if we can cook up something like that. Since perception is all that exists, our theory of how the universe and everything in it works is necessarily limited to notion of 'causes' imposed by fiat onto correlations. Again, no way out.
--
6)Taking human brain as the 'cause' of mind, or at the very least, as a reliable, consistent correlate of the activity of the mind, we see that all grey matter is pretty similar and in functional terms, do the same thing. Molecules(Na/K/5-HT..etc) move from one place to another and back. There are a bunch of rules on how this occurs. However, this activity is homogenous across the grey matter. This grey matter gives rise to both emotions and cognition as well as the different sensory modalities as well as motor control. So, from reverse mapping, emotions follow the same 'logic' as rational thought. The functional atoms of emotions may be different from those of cognition, but the rules of mechanics for both set of atoms is the same. So, AI as logic, needs to figure out how to objectify emotions and basically all neural activity into a codified framework of human psychoactivity. I don't think the core approach is itself wrong. The domain might be.

posted by Gyan at 7:31 PM on November 22, 2004

John Waters also said. "Never sleep with cast, crew is better."

As a stagehand, I will not dispute this.
posted by oaf at 9:18 PM on November 22, 2004

This article inspired me to join, finally. Anyway, I'm a current student in the Minds & Machines program under Selmer, and I think it's interesting to note that Selmer's main purpose in creating Brutus was to show how bad Brutus is at writing stories. At least, that's what he said the last time he lectured about it. Selmer's an interesting guy, he believes that the pursuit of AI will be entirely fruitless, but he works on it anyway. Basically, he's a full dualist and doesnt think that AI will ever be able to match humans. He supports logic in AI because he thinks it's a good problem solving strategy, not because humans are that way. Ryvar, his views very well could have changed in the last 4 years, but these days he doesn't think that humans are logical in the least, and regularly lectures on why normal humans are inferior to him because of his superior logical abilities.
posted by JZig at 9:51 PM on November 22, 2004

The RIAA is probably funding research for a similar project to generate "musicproduct" for sale without having to deal with those pesky musicians...
posted by Enron Hubbard at 6:42 AM on November 23, 2004

Opening line of Ha Jin's Waiting
Every summer Lin kong returned to Goose Village to divorce his wife, Shuyu.

In your face logic!
posted by TheSpook at 12:29 PM on November 23, 2004

Very interesting, JZig. Thanks.
posted by semmi at 8:12 PM on November 24, 2004