How To Break Anything

Thoughts and insights on culture and human behavior, living blissfully at the intersection of rationality and irrationality (but mostly irrationality) 
Filed under

intelligence

 

Communicating with robots, and a few thoughts on the complexity of communication

New York Time's Interview With A Robot (video here, no embed code) gives an interesting look at how we think about interaction and conversation. Much of the commentary around the video is on it's "disturbing" nature, reminding me of the idea I ran into once that if something attempts to appear human, but falls short of being 100%, we find it to be disturbing. 

I recommend watching it. The video got me interested again in an old hobby of mine: conversing with the Jabberywacky artificial intelligence bot. Evidently the creators have now built the similar Cleverbot. I had a couple conversations with it below (the first couple of lines were cut off from the screenshot):

Kyle: Hi Cleverbot, let's talk about what it means to be human

Cleverbot: I know I am human.

Cleverbot got a little distracted at that point so I decided to start again:

Clearly, as in the above video, one of the problems we're still struggling with in terms of recreating intelligent conversation is understanding how humans react not to just individual fragments of conversations but maintain a memory of the context of entire conversations. As you might imagine this is no simple task; it requires a deceptively complex understanding of how memory works, in allowing us to relate just the right amount of information now to just the right amount of information from the past.
 
A good way to think of how this problem is exponentially complex is to think of the potential semiotic significance of one statement as limited to let's say just 5 different meanings. The potential expression is much more difficult to pin down when considering it within the context of an earlier statement with another five potential meanings, and another statement before that. Conversation is clearly much more than an exercise in reacting to the last thing stated.
 
(As an aside, I found it interesting to note that Cleverbot responds back in deliberately slow typed characters, simulating human typing. Nice touch.)

Loading mentions Retweet
Filed under  //   behavior   intelligence  

Comments [0]

Danah Boyd on ChatRoulette, and myself on exposure and intelligence

I feel pretty depressed every time I watch people flip out about the dangers of talking to strangers. Strangers helped me become who I was. Strangers taught me about a different world than what I knew in my small town. Strangers allowed me to see from a different perspective. Strangers introduced me to academia, gender theory, Ivy League colleges, the politics of war, etc. So I hate how we vilify all strangers as inherently bad. Did I meet some sketchballs on the Internet when I was a teen? DEFINITELY. They were weird; I moved on.

Danah Boyd on ChatRoulette, above.

In addition to being about control of our identity, privacy is about protecting ourselves from strangers.

To which I have a similar reaction as Danah's here:

"I’m still not sure what to say except that I feel this weighted sense of Le Sigh. The same mix of depression and exhaustion I felt this morning when I was playing peek-a-boo with a smiley child in an airport and her parents whisked her away, glaring at me as though I was the devil incarnate. I realize that many parents think that they’re doing good by their kids when they choose to limit their exposure to the randomness of the world, but it just makes me deeply deeply sad."

I still vividly remember once upon a time when I was little, when I was watching Interview With A Vampire with my parents. There was a family friend there as well. I couldn't have been more than 10 years old. As I remember, there's a part where one of the female characters is nude for some reason (If I remember correctly it's part of some gruesome performance where she's devoured by vampires).

I don't remember the exact comments, but I remember the family friend making some objection about me being there (not because of the gruesome scene, but because of the nudity), and I remember my parents making some response about exposure and reality.

And I remember it because it so deeply reflects the world I grew up in, in which exposure to the world mattered above all, for all it's grittiness and for all it's conflicting, differing, and paradoxical views that constantly change the way I think about things.

At some point I'll explain the idea of defining intelligence as "efficient cross-domain optimization," as I ran into at this weekend's NYC Future Salon. But for now, suffice to say that this exposure has made all the difference to me in the cross-domain realm, in the same sense that Danah expresses above: were it not for exposure to weird and scary strangers, it's likely she'd never be the academic she is today.

Loading mentions Retweet
Filed under  //   control   intelligence   perspective   privacy  

Comments [0]

My Blogroll: