No story about AI can be complete unless one has cosidered: The Emperor's New Mind, and also: Shadows of the Mind, both by Roger Penrose, erstwhile collaborator with Stephen Hawking. He gives the most balanced and factual examination of AI and whether any combination of algorithms could result in non-preprogrammed intelligence.

PS, Aman, Creativity Machines consist of computer GENERATORS AND CONSUMERS. It's just that they are typically joined at the hip, so to speak, on the same computer (see http://imagination-engines.com/iei_basics.htm). It makes no difference if they are separated and on different computers.

With all due respect, Aman, this is from http://www.popmatters.com/pm/article/computers-compose-personalized-music/

"Thaler sat in front of the Web cam and listened. He indicated his likes with a smile and displeasure with a frown. A critic program connected to the music-generating program learned by watching the emotions play on his face what Thaler liked and gave him more of that. It learned what he didn’t like and gave him less of that."

Here is the self-editing, self-curating you require. The system tailored itself to its one man audience and was quite capable of scaling to a packed concert hall...QED!

For the bleeding hearts, it should be mentioned that these systems utilized a replication of the verty neural circuitry used in the brain to produce subjective experience and emotion.

The PRI program omitted many juicy details, such as the major paradigm shift the Creativity Machine technology brought to the field of neural nets, major news coverage in Scientific American, New Scientist, and a proclamation from NASA that this is the sole technology that can enable human to trans-human level intelligence in machines, not some programmers writing "if-then-else."

...I'm sure you're a great performing artist though.

I am a professional performing artist. I really appreciate the piece and the pointers to this work. It's fascinating.

However, a fancy tool is not an artist. These computers or programs are sophisticated artist's TOOLS. You might even call them 'instruments'. They're not artists.

Here's the deal: to be called an artist, the program or computer must be able, at least in part, to decide WHAT to present to the public. The self-editing and self-curating process is essential to artistry. No esthetic judgment?—No artist! No capability to judge others' art?—No artist!

The ability to surprise is REQUIRED, but it is not is not enough. The capability to surprise—even to surprise the builder of the software or computer—is not sufficient for artistry.

Neural nets aren't new. Rule-based expert systems aren't new. This particular LEVEL of work might be new. But I bought a hard-bound book of AI-produced art back in the 1980s.

Finally, as a performing artist myself, I disagree with Chris Gurin that the artist-software needs to *feel* a certain way to be an artist. Feeling a certain way is neither necessary nor sufficient for being an artist.

I DO believe we will have computer artists someday. But first we'll have computer READERS and CONSUMERS of art and literature.

There is no artist without an esthetic.

I agree with sploosh, but I find it ironic that everyone is fascinated with what is effectively the "hand-crafting" of a Rolls Royce. Of course there are no surprises for those toiling away in the factory since the vehicle will function according to spec. The plant will turn out a finished luxury car that will have great appeal and mystique to its new owner who has not been jaded by the day-in-day-out manufacturing process.

I am admittedly more impressed by the Rolls that suddenly assembles itself and if need be transforms into a yacht, a helicopter, or ocean liner, on an as needed basis. That is what this gentleman is talking about regarding the creative machine. It can be anything, on its own!

This all makes me wonder if computer scientists are really scientists, because institutions like RPI should, by definition, be scientifically curious, especially about a technology that can easily do everything discussed in this article and vastly more. Perhaps this is a case of so-called scientific institutions behaving like a prideful businesses and a business acting more like a scientific enterprise.

shouldn't the "dingularity" music be prefaced by "self-written by machine intelligence" rather than "programmed by stephen thaler"? i think someone there seriously missed the point.

see http://www.ericleech.com/2010/12/2012-is-the-end-really-near/ for truly machine composed literature.

are we going to be hearing about what's new at the rpi student union next?

I have a class with Selmer this spring. I will be picking his brain about this, because this is pretty cool.

Oh, one of my favorite kind of Studio360 shows! Focusing on technology and creativity! Thank you!

And it opens with a segment on RPI (my PhD alma mater) and Selmer Bringsjord. Wow, he's come a long way since working with his early partner on the "Autopoeisis" project, with David Porush (the actual "literature" professor who collaborated with Bringsjord in the early 90s, when I was there, and my dissertation advisor).

And he's running his own department now. Neato. Used to be housed in the Dept of Philosophy, which was an odd place for folks working in AI, which cognitive psych was as well).

David is no longer at RPI, so maybe that's why he wasn't mentioned. I wasn't sure Bringsjord was still working on AI fiction projects. Gonna look into it again now.

Thanks for the blast from the past, and on my favorite show too! (I really enjoyed attending the Taboo program live in the Greene Space, btw-- the program was just hilarious).

"Art is the child of time and of percedent. It inherits the ages; but unless the artist comes into his inheritance, he is helpless. As best, can he go but one little step beyond the fathers, add one little stone to the edivace; and inorder to accomplish even this much, he must know well the work of his predecessors. If by some dreadful catastrophe all the art of the world should suddenly be destroyed and all knowledge of it be blotted from theh minds of the survivors, it would require ten thousand years for humanity to recover the lost ground."
Landscape paining by Birge Harrison

true art evokes emotion. machines have no emotion... yet.

@Julia from Baltimore
My thoughts exactly. I would go on to say that our perceptions of art are how we give it its context, especially when we can't communicate with the artist. Think of the Caves of Lascaux- if anything, it's our contemporary interpretations of the ancient cave paintings that are creative.

The article spoke much about ideas or stories that we create in our minds to make sense- contextualize- the art. The example of the WNYC art critic telling an interesting story of the visual image created by the computer; the stories that we impute to the computer generated film. These are all examples of the viewer, the consumer of art, creating.

This leads to the idea of the mind-sense of buddhism. We do perceive reality through our five senses, but we also seem to create an understanding of that reality with our sixth, mind-sense.

After all, think of how the WNYC art critic would've responded had she known the drawing was made by a computer?

The real question is not, can computers create art but "what is art?" I don't believe there is any general consensus on that, is there? According to some definitions, yes they can, either now or in the future. But, according to some factors which seem essential in my mind, no they cannot. One being simply that, for me, art is as much about the process as the product. The exquisite enjoyment of and engagement with the 'work' or making art is its essence for me. Even if computers could theoretically have that experience, how would we ever know?

Secondly, the story behind a work of art is as important for me as the isolated piece. In fact, I would venture to say that no piece or art is ever isolated from its context, personal, economic, political (including the politics of the business of art), historical, etc. I suppose there is a context to computer art in the sense that computers are a part of society now. But I don't find that a particularly meaningful attraction.

I just happened to have listened to a very interesting and pertinent lecture on the neurology of art by the well-known V. Ramachandran. Here's the link: http://www.bbc.co.uk/radio4/reith2003/lecture3.shtml

The segment on WHITE ON WHITE reminds me of an upcoming study of Film Noir from Shannon Clute and Richard Edwards, The Maltese Touch of Evil. Highly recommended.

C'mon guys...I kept waiting for it.
Keats;
'Beauty is truth, truth beauty,—that is all
Ye know on earth, and all ye need to know.'

As a software engineer with some AI experience, I know that the best programs are those that 'teach' themselves, something I wish my GPS could do. When I continually override what it tells me, it should incorporate my chosen route into its mapping. Even put a switch in it. I could screw this computer up by giving it meanings opposite to those we generally accept (e.g., 'dark' for the word 'light'; or 'illuminating' , a little more subtle, for the word 'dark').

"Are Computers Creative?"
The part of creativity no computer, or algorithm, can implement yet is the emotional feedback we experience from others and from ourselves. As a student, I'm compelled to be creative on a deadline, to strip away all the extraneous bits, and focus on essential criteria set by my professors (that is, if I want to avoid pulling too many "all-nighters"). This can often be an emotional roller coaster: I can see the end of my imagined efforts fully formed in my mind, and anticipate the pleasure of the realized vision, but then must deal with the reality of limited time and still evolving skill. There are times I've followed my emotional arc to success, but more often to a blank wall, then had to back-track to essentials and no more. That hurts, but being creative often means learning to, as one novelist put it, "eat your babies." When I've finally got it right, found a satisfying compromise between my desire and my deadline, I "feel it" in my guts. Unless and until a computer can "feel" (now, define "feel"), it can do no more than implement rule-based approaches: competent, but lifeless.

What was the movie they were talking about? The one with the algorithm.

Both approaches are neat, but the neural solution is capable of so much more.

Great piece!