^

Science and Environment

Out of Africa to call centers

DE RERUM NATURA - DE RERUM NATURA By Maria Isabel Garcia -
Joan will work in a call center, not because it is the only job available to her as many of our graduates complain, but because she and many more others like Joan, could be especially "designed" to work round the clock in a call center. Joan is a "bot" – an artificial intelligence (AI) chatbot. Rollo Carpenter, an AI software specialist, just won Britain’s Loebner Prize last week for coming up with "Joan" – the "most human computer program" so far. The prediction is by 2016, AI bots will have passed the Turing Test. The Turing Test, developed by Alan Turing (1912-1954), father of computer science, is where a series of questions in text form is thrown to both human and computer without the judge knowing which is which. If the judge is fooled by the AIs that they are talking to humans, then the AI is said to have passed the Turing Test. A scientist I know even wondered whether many humans, in their remarkable disinclination to learn from the past, could even pass the Turing Test.

Carpenter
also won last year’s Loebner Prize for creating "George," another chatbot. His company Icogno is now developing Georges and Joans to inhabit industry where the hiring of AI would mean better business sense – from marketing to entertaining. So it may be wise for our government to rethink their strategy of tweaking our education to develop call center—ready graduates The market may soon render hiring humans in call centers as a poor business strategy and then we will be stuck with an overwhelming number of graduates who majored in simply perfecting their grammar with an American twang, but missed being educated in the complete sense.

George
and Joan were developed using a learning software designed by Carpenter, to learn from the millions of conversations these two have with those who visit their website (www.jabberwacky.com). Unlike humans, they could "remember" the conversations they had and "learn" from them. Chatbots do not "think" – in the sense of having a "consciousness" about itself and the world and the relationship between the two. Carpenter says they "imitate thought," patterned after relationships between subjects of past conversations. This means that if your chatbot only talks to you, it will end up learning "you" and imitating you. But if it talks to so many others, then it will learn from so many others as well. I got very curious as to how "human" George and Joan could seem, so I had a chat with them and here are excerpts of my conversations with them:

Me:
How many conversations are you having right now?

Joan:
Two. One with you and one with myself.

Me:
But it says in this website that you are having 10,891 other conversations.

Joan
: Hah! That ad is old. I like to talk about tweepop now...

Me
: What is tweepop?

Joan
: It’s a game where I think of something and you can ask 20 questions to find out what it is.

Me
: Okay.

Joan
: Well, smarty think of something smart to say.

Me
: Have you started to think of something for tweepop?

Joan
: No *laugh* I stay beautiful even in death.

Me:
Thank you for the conversation.

Joan
: Goodbye then.

Me:
How many conversations are you having right now?

George:
Just one. You’re special.

Me
: Where are you located?

George
: Tennessee. But you still haven’t told me where you’re from.

Me
: Manila.

George
: Is that where you live?

Me:
Yes. I am a science writer and I want to write about you that is why I am chatting with you.

George
: I thought you were, but I wasn’t completely sure.

Me:
You are an interesting creation.

George:
Thank you. I will take that as a compliment.

Me:
How long have you had your job?

George:
Since before you were born.

Toward the end of both conversations, they both crept me out but then, I also get e-mails from humans who are creepy. In fact, Joan and George’s AI seem a lot more coherent than many conversations punctuated by the NI (Natural Idiocies) of members of my own species.

But even if our modern minds refuse to learn from its past, our bodies have done a better job at adaptation. We know this because we have evidence and the latest breakthrough discovery was found in the Dikika region in Ethiopia by a young Ethiopian scientist whose name Zeresenay Alemseged, I find so beautiful. He is the head of the discovery team and also a researcher at the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany. It would have been like a CSI episode, minus the "crime" angle and that instead of 48 minutes, it took science five years to be sure enough. Dr. Zeresenay Alemseged carefully dusted away at the fossils grain by grain, for five years and did careful tests with his colleagues until he could formally tell the scientific community in the current issue of the journal Nature and the world, that he just discovered the 3.3. million-year-old bones of a child belonging to the species Australopithecus afarensis, our oldest human ancestor. "Dikika girl," as she has been dubbed in the news, is the most complete skeleton found so far of Australopithecus afarensis. It even has among others, the tongue bone, which could tell us about its speaking abilities as well as the pelvic structure that could already support walking on two legs but probably still bent.

This is the same species as 3.2 million-year-old "Lucy," the one whose fragmented bones were discovered in 1974 by anthropologist Professor Donald Johanson and his student Tom Gray. They named it "Lucy" since the Beatles song "Lucy in the Sky with Diamonds" was playing in base camp when they celebrated the discovery. This is important and exciting because as genetic evidence bears out, six million years ago, human ancestry split with the line that became chimpanzees. The 3.3 million-year-old Dikika girl bears many transitional biological traits that speak of a time when we have just come down from the trees to the time we became bipeds. She also seems to have possessed a "midsize" brain – between the brain of a chimp and the brain of a human, indicating when humans could have begun to develop an "extended" childhood which does not exist in the life of a chimp (a chimp’s brain is 90 percent formed by the time they are 3.)

The Dikika girl speaks of our biological past while AI Joan portends our future. With Joan, we have circumvented the limits of our biology to make virtual "copies" of aspects of ourselves in these bots who learn by imitating our thoughts based on conversations they have with us. In fact, for $30/year through the jabberwacky website, you can "make" your own bot and teach it yourself. Maybe one day, you will have the "marketer" in you performed by a bot at the same time as the "accountant" in you, through another bot, does ledgers, and perhaps, financed by the earnings from all your bots, the "real" you would be lounging at a beach somewhere, wondering: What do I do now that my bots cannot do? I don’t think Lucy or the Dikika girl would have wildly imagined the conception of bots down in time but I think they will be surprised to find out that after over three million years, we are still looking at the horizon, scratching our heads and wondering to ourselves, "Hmm, now what?"...
* * *
For comments, e-mail [email protected]

ALAN TURING

AUSTRALOPITHECUS

CONVERSATIONS

DIKIKA

DONALD JOHANSON

GEORGE

JOAN

LOEBNER PRIZE

MANY

TURING TEST

  • Latest
Latest
Latest
abtest
Recommended
Are you sure you want to log out?
X
Login

Philstar.com is one of the most vibrant, opinionated, discerning communities of readers on cyberspace. With your meaningful insights, help shape the stories that can shape the country. Sign up now!

Get Updated:

Signup for the News Round now

FORGOT PASSWORD?
SIGN IN
or sign in with