Specific comments:
It is interesting that his opinion is "The idea of 'intelligent agents' is both wrong and evil." Not the "research" or the "creation," but the "idea" itself is evil.
Problems with his practical problems:
As Lanier describes how "agents make people redefine themselves into lesser beings," several of the steps seem to be exaggerations of Asimovian proportions (in one Asimov story, society forgets math because calculators do it all):
His rhetorical "technique" of casting doubt on a statement by questioning the author's sanity is actually parodied in a popular humorous document on the internet, "How to Flame." (for example: see The 12 Commandments of Flaming (http://www.blakjak.demon.co.uk/flaming.htm), ). This implication immediately follows an accusation that Turing made his argument because it was "in vogue." (Ironically, this general argument seems a rather trendy one to make.)
That Lanier includes as evidence something that he admits is false in a footnote is bad form and confusing. Interestingly, he switches to the passive voice in the footnote, claiming "that is the way it is taught," as if to impute more authority to the statement. Indeed, I never learned it the way he claims it is taught.
Lanier's argument that there is a problem with a Turing Test because it can't compare the way humans are NOW to the way computer will be in a future seems flimsy and pointless to me.
I like the statement: "The artificial intelligence question is the abortion question of the compuer world." It seems very accurate.
Again, he resorts to psychoanalysis of an opponent as opposed to showing why the actual claim is false: "Part of the reason for [my opponent's arguments] might be that many in the computer world are attracted to the deathless world of abstraction, and nurture hopes of being able to live forever by backing htemselves onto a computer tape."
General comments:
I agree with his argument that widespread use of poor-quality agents creates a viscious cycle, by which people come to expect less, and agents never get any better, but I do not think the situation is nearly as dire as he predicts. Bad agents will inspire people not to use them (as it obviously does today), and will not cause people to anthropomorphize them and therefore limit their own potential. I have watched my techno-savvy, Wired reading friends experiment with agents, realize their limited usefullness, and stop using them. Lanier's views of humans as naive children are overly pessimistic.Specific comments:
I strongly agree that previous proponents of artificial intelligence have been misled by their own rhetoric.
General comments:
It is interesting that the textbook "Artificial Intelligence: A modern approach" (Russell and Norvig), the canonical AI text, is organized around the goal of creating an atonomous intelligent agent.
Specific comments:
The figure is confusing. Why are the ususal circles of the Venn diagram not used?
Nadin makes the claim that semiotic considerations must be integrated at the level of the operating system; however, no specifics are given, and I have NO idea how this is possible or useful.
General comments:
This seems (to me -- a newcomer to the field) like a very well put-together essay by someone trying to push the field of HCI back on course. I would like more elaboration on how the increasing complexity of computers and their programs necessitates a more formal semiotic foundation. I would also like to see examples of what a formal semiotic foundation looks like.
To list of readings
To contents
To contents