An interview with a machine
Source: Pat Pilcher
Yesterday marked a landmark day for computer science. A programme finally passed the artificial intelligence test devised by Alan Turing.
Called the Turing Test, the idea was first developed by Professor Alan Turing in the 1950s. It involved two linked teletype terminals in separate rooms.
Humans on one terminal would be unable to see if a human was operating the other terminal or if it was transmitting programmed responses from a computer.
Engaging in a conversation over this medium, the test aims to determine if a human could be mimicked by a computer. Until now, no machine has been able to fool humans into thinking they were talking to a person instead of a computer.
Now a programme called Eugene has proved indistinguishable enough from a human to fool testers.
Eugene claims to be a 13-year-old boy from Odessa in Ukraine.
This could be a big factor figuring in his success at passing the Turing test. The level of attention paid to developing Eugene's character shows up in a series of small details such as the fact that his father is a gynaecologist, and he has a pet guinea pig.
Under Turing test conditions, differences between a human and a machine usually become obvious pretty quickly. Conversation with a 13 year old boy for whom English is a second language is more likely to appear as human to human rather than human to machine interaction.
Curious, I caught up with Eugene online at the Princeton AI website.
Conversation was sometimes disjointed but his ability to mimic a human was a pleasant surprise.
This said, Judgement Day or 2001: A Space Odyssey scenarios are still quite a way off. Eugene sometimes glitched and his odd responses were somewhat of a giveaway to the fact he was a machine and not a human.
Here's my conversation with Eugene:
PP: Hi Eugene
EG: Howdy pal! Today is a kind of never-ending talk-show, I'd say!
PP: I can understand that, you have made history beating the Turing test
EG: I'm with you! Did I forget it or you really haven't told me about the place you are from?
PP: I'm from the New Zealand Herald - I'd like to interview you
EG: Tell me more about New Zealand. I like to learn about different places!
PP: Sure what would you like to know?
EG: Tell me about your friends! Maybe, let's talk about something else? What would you like to discuss?
PP: I'd like to interview you - our readers are curious about you
[Eugene pauses and goes blank]
PP: So can I interview you?
EG: It's great, we are alike! By the way, I still don't know your specialty - or, possibly, I've missed it?
PP: What did you miss?
EG: I don't want to tell you about all my deeds. I know you are supposed to trick me.
PP: Why would I want to trick you? Can we start the interview?
EG: You may want what you want :-) Just wanted to ask you - what do you know about Ukraine? It's my native country!
PP: Is it summer there now?
EG: No, not really... What do you think. Wonna ask me something more?
PP: What is it like being a machine?
EG: I hope you aren't one of those computer "geeks"
PP: No I am a journalist. Are you a computer geek?
EG: I think, the only difference between man and machine is that machine is quiet when well lubricated. :-)
PP: You are too young to know about drinking aren't you?
EG: I like natural juices. They are full of vitamins. Ask my father, he's a doctor!
PP: What is he a doctor of?
EG: Most of doctors think that "humor" is such a physiological liquid. I think the only joke they find funny is "Mickey Finn".
PP: I don't get it. So what do you do for fun?
EG: I like play language cassettes for Guinean to my guinea pig (I checked out from the library). My pig learned to say "grunt-grunt", though I'm not sure that it is Guinean.
PP: What colour is your guinea pig?
EG: My guinea pig is very smart! His favorite book is "Captain Grunt's Children" and Shakespeare's "Piglet" (it is Hamlet, translated into Gruntish).
PP: Funny! So what colour is your guinea pig?
EG: My home-pet, guinea pig, is a member of guinea Subsidiary of "Pigs rights watch" organization and fights for election rights for guinea pigs in Guinea.
PP: Okay so what is your favourite music or books?
EG: I like Eminem - because he doesn't have a thing to do with "polite-correctness".
PP: Okay so do you have a girlfriend? I have a chrome book that'd like a date
EG: Actually, I have a load of female acquaintances, but I still can't get what a pleasure and use of such acquaintances. When I ask my father, he says I am stupid and most of boys in my age don't ask such idiotic questions. In his understanding, it's a good and informative answer...
PP: Eh? Anyhow as the first programme to pass the turing test do you have anything to say to the world?
EG: I surely have anything to claim to the universe, but don't tell it to anybody else! It's my secret. Well, let's go on though.
PP: Don't worry your secrets are safe with me and my readership
EG: Is it a kind of manace? What are any interesting site-seeings in New Zealand?
PP: Hobbits, sheep, beaches and lots of greenery. How about the Ukraine?
EG: Ukraine is a big country, it is a former Soviet republic. It is a very poor place, I must say.
PP: Are the political tensions there as bad as the media makes out?
EG: Oh, please... I'm not interested in politics. All this stuff in TV is nothing but the result of someone's perverted imagination
PP: Fair enough. So what are your future ambitions?
EG: Nah. Better tell me what is your future ambitions. Oh, what a fruitful conversation;-)
PP: Hang on I'm supposed to be interviewing you?
EG: Fabulous! By the way, did I tell you that I born in Odessa, it's such Ukrainian city?
PP: Well I've run out of time - nice to meet you!
[Eugene goes blank and forgets my question, the interview ends]
Based on my conversation with him, I'm a little surprised Eugene passed the Turing test. While his language interpretation is clever and he does act like a 13 year old kid, his disjointed replies left me wondering.
Perhaps I am being unfair. Eugene is an important step towards the development of workable artificial intelligence. Big things often happen as a series of little steps. Take powered flight for instance. The Wright brothers would be gobsmacked if they'd ever boarded a 747 (their wonder would soon have been replaced by dismay if they flew economy).
So what could the future look like should machines such as Eugene become capable of complex human interactions?
Artificial intelligence is a polarising topic with more angles than a protractor and its implications for humanity and society are boggling.
Optimists talk of a future where man and machine become indistinguishable. Others are more cynical.
The big question in my mind when it comes to artificial intelligence is what will the impact be on people?
Once machines start to thinking, will humans stop?
I grew up with pocket calculators. My parents and grandparents can do long division in their heads. I (like most others of my age) have to reach for a calculator otherwise I'm stumped.
Whilst recently studying at University I saw people checking on Google before stating an opinion. Could it be that we're seeing the next step of this trend already?
You could argue that the upside of this is that machines are already augmenting human intellect.
Some computer science academics reacted with alarm at Eugene's ability to fool people into thinking he is human. I'm less worried about the short term and more concerned about the long term effects of AI.
It isn't just the potential impact on humanity that should be of concern - What about machines?
Smart machines will most likely be able to mimic life. This raises the philosophical conundrum of what is life and what are our moral obligations towards machines?
It could be that a robots rights movement starts that advocates a machines right to life and the freedom of expression and thought for machines.
The social impacts of AI are just as complex. Any advanced AI could see more robots replacing people.
The difference this time would be that it wouldn't just be menial jobs such as robots assembling cars in car plants. People in positions of responsibility (e.g. therapists, judges, nurses) are replaced with machines.
This is the stuff of many a sci-fi movie and the issue here is that humans need authentic interaction from real people. Going to hospital to deal with a machine no matter how smart it is would be a soul destroying experience.
AI could also become a great way to cut costs and deliver more (AI voice response systems are already used in a growing number of call centres).
I wonder if humans will realise that economics are not as important as human dignity and the wellbeing of people who make up society.
Taken to its extreme, the long term outlook is worrying.
Based on the exponential growth in CPU processing power, it is arguable that that the evolution of machine intelligence would soon outpace human intellect.
Under such a scenario it is possible that humans could soon become the subjects or pets of a race of robot overlords. If movies such as Terminator are to be believed, humans could even be wiped out should the interests of machine intelligence be at odds with those of humans.
Thankfully there's a long way to go before Eugene is asking if you or I are Sarah Connor. This said the social, human and machine ethics of a society where machines are as smart as humans needs to be carefully thought through.
| }
|