how big a database of text do we need to store, in order to have a bot which passes the Turing test?
In fact, even Eliza has passed the Turing Test for some humans - it was reported in the 1960’s that people interacting with it thought they were talking to a real person.
I played with the Eliza program in the 1980’s - even though I knew it was a machine, I found the conversations could be interesting.
Given enough templates stored in Alice, I think it could pass the Turing Test. But how many templates would we need? a million? a billion? a billion billion?]]>
Do you know of any chatbots which play on the concept of Ouija Boards, Magic 8 Balls, or other forms of divination? It strikes me that divination conversations are *supposed* to be partly incoherent… however the normal Eliza rhetorical trick of “no, tell me more about you” won’t work very well with a divination chatbot.]]>
Nice import with the Goth Disco bot scene and then Harry Potter’s Tom Riddle Diary (the later you beat me on). I’ll see you with four Narnia bots: Aravis, Puddleglum, Reepicheep and Mr Tumnus. There is a relationship here, one that follows on from discussions about adaptation. In a sense, we have the same material content (source code and AIML) that is reframed, repurposed. The source code is only implied (see Jeremy’s “implied code” maybe here?) and the AIML is evidenced once a conversation takes place. But before the conversation there are many cues, many framing devices that seek to persuade, to affect our reading of the AIML.
Usually, in debates about the nature of adaptation (as far as I know) there is not question as to how the difference between peices because they have been created mostly anew. Pages from a book, for instance, don’t become the celluloid in a film. But with the age of “convergence” we’re having to rethink our meanings of new (and by convergence I’m referring to a particular definition of data that can be reused in many different media, for instance: a webpage that can be viewed on a computer, PDA and mobile phone). The idea that a program can be utilised by many is a given, but the idea that that program has a role in the creative process both at the point of creation and reception is a bit newer. Then we have the problem of code within code, the AIML, the natural language that links the program to the human. If the reuse of this part of a program is a problem/issue then it follows that we’re assessing it on different grounds. Is the AIML a unique expression of an artist (ie: Richard Wallace and the many botmasters who have contributed to it) or is it something else? At what point in the botmastering process does the bot become a new artistic creation? 100 patterns? 10,000 patterns? Never?
And yes, I find it a funny phenomenon that bots can be reframed in so many ways with relative success. Is this unique to bots? I mean, we’re all familiar with copycat works but this is going another step.
Another couple of things:
1) I like the development of fan fiction to fanbots. Fans have focused their creative energies on exploring their favourite characters in settings not coverd by the original author or not intented. Now they’re going that step further and creating them.
2) Fanbots falls nicely in the Narrative Universe model where the content is unsanctioned but is still within as storyworld. It shows how audiences want to persist a storyworld no matter what. The boundary of a storyworld is no-longer governed by the original or commissioned creators, although its rules are mainly seeded by them.
3) And yes, I like the film cuts in the goth sequence. Nice to have a change to the bot sequences. I felt it was moving into film, game and SIMS territory, perhaps too much…I like the directness of a bot address. It is a completely different experience to the game experience.
4) On the topic of a sweet-spot, could we consider a bot is more likely to be liked (in light of the context framing that is so popular) if it utilises the following:
a) a well-known fantasy character;
b) a therapist
c) a superhuman (God, angel, devil)…
Drawing as it does on “Harry Potter and the Chamber of Secrets,” Tom Riddle’s Diary by Icegreen Technologies provides another interesting context, as long as we’re on the topic. This is fun because it takes the magical pre-technology of the book and film and remediates it in a contemporary electronic form.]]>
I’d also like to point out another mostly Out of the Box (OOTB) ALICEbot that used on another site linked from the alice page.
Go to this page and select #26.
What strikes me about this particular Flash application is that the addition of changing camera perspectives (albeit animated ones) creates an additional layer of presence that is not present in the text-only versions. [If you carry on your conversation with the woman for several times, the camera angle will change ala shot/reverse-shot (SRS)].
What I’m thinking is that maybe one way to categorize some of these bot projects is in how pre-existing bots are recontextualized and repurposed. Something tells me that of the thousands of ALICEbots and other customizable bots that there are more than just a few of these. But this is not to criticize, so much as to point out that recontextualization is a kind of creative modification of the bot. If we go back to Turing, it is as if these provide new thought-experiments or perhaps better new metaphors in which to convey this technology.
I also like the idea that there is a “sweet spot” but, again, I see it in the conceit. ELIZA is the original and still one of the best concepts for the bot. The deranged or forgetful people our other interesitng ones. So in addition to the uses of the technology, there is the all-important concept, I’d say, that informs one branch of bot development in the lineage of ELIZA.]]>