John Markoff in the NY Times writes about a new generation of AI researchers, among them a Robert Hecht-Nielsen:
"Last year he began speaking publicly about his theory of “confabulation,” a hypothesis about the way the brain makes decisions. At a recent I.B.M. symposium, Mr. Hecht-Nielsen showed off a model of confabulation, demonstrating how his software program could read two sentences from The Detroit Free Press and create a third sentence that both made sense and was a natural extension of the previous text.
For example, the program read: “He started his goodbyes with a morning audience with Queen Elizabeth II at Buckingham Palace, sharing coffee, tea, cookies and his desire for a golf rematch with her son, Prince Andrew. The visit came after Clinton made the rounds through Ireland and Northern Ireland to offer support for the flagging peace process there.”
The program then generated a sentence that read: “The two leaders also discussed bilateral cooperation in various fields.”
(http://www.nytimes.com/2006/07/18/technology/18brain.html)
Elsewhere we learn that Hecht-Nielsen’s program has already “read” 8,000 encyclopedias and several thousand novels, processing relative word frequencies and sentence content and structure. An interesting spin on the Turing test: rather than a machine that, communicating through the disembodied medium of text, can answer questions and proffer statements, we have instead a machine that can confabulate, produce, add to an already given body of text. Turing envisioned a test in which the irreducible quantum of humanity would be conversation, but here it is pure production.
The example chosen is particularly menacing, as it shows how rigorously mechanical our own informational economy is. “The two leaders also discussed bilateral cooperation:” of course this statement is a pure confabulation, a fictive mechanical extrapolation from a factual reportage; yet it is indistinguishable in fact from any human-created press release. We find ourselves quite suddenly at a horizon of meaninglessness, at a place where facticity recedes as something expected and offered. There is nothing mechanical about confabulation—or rather, nothing inhuman about it. Perhaps we could tell if it were really a computer behind the curtain if we were to ask it about its children, but could we ever judge between a press release created by a computer and one created by a human? And, more relevantly, would we want to?
The true apotheosis of artificial intelligence will occur not at the point where we cannot differentiate between man and machine—after all, there will always be more and stricter tests to implement—but the point at which we no longer care to try. Only when a machine can disappear into the world of text, when it can occupy the empty, repetitive, mechanical space inherent in the human, will we accept it as one of us.