Here we are pleased to present an interview with Mohan Embar the winner of this year's prize with his chat bot Chip Vivant. Congratulations to Mohan on behalf of all the members here at AiDreams. And without further ado...
It felt wonderful, obviously. Although the first time I entered was in 2008, I had been dreaming about writing a chatbot since 2000 or so and actually had an interest in chatbots since my childhood. The win was all-the-more meaningful because I won on my terms. I'll touch more on this later.
With the exception of a six-month period in late 2007 / early 2008, it's been in fits and spurts. Probably a few weeks part time before each competition.
"...the thing that pleased me the most is that I won on my terms, without a fake backstory, fake typing errors and a preponderance of canned responses."
That's also the reason I feel so thrilled about this win. As described further on that page, on the day of the win, I shared this message with my friends at Robitron and chatbots.org:
People - I'm still in shock with what just happened here. One of my main criticisms of this contest is that despite Hugh's vision, the judges' slavish interpretation of the contest rules favored fake backstories, canned responses and other trickery over real effort.
Well, Chip won:
This flies in the face of a lot of long-cherished beliefs people have about this contest, including my own. I don't know if it's a fluke, but all of the judges were pretty consistent in how they approached this.
The above doesn't mean that Chip doesn't have his fair share of canned responses. I personally abhor canned responses, though. (I hired a contractor to write a bunch of them for me for this year's contest because I couldn't bring myself to do this myself.) I mention my reasons why here:
Programming a bot to pretend to be a human involves much more than one line of code where the bot affirms that it's human - it involves an extremely labor-intensive (and IMO time-wasting) effort to code up a web of lies which invariably implodes under its own weight.
The fact that Chip won means that despite the naysayers, the Loebner Prize Competition is a legitimate tool for attempting to advance the field of AI and not "an obnoxious and unproductive annual publicity campaign" as Marvin Minsky put it.
I think Cleverbot would probably win the competition if it could make it past the prescreening round, which it usually doesn't because Hugh's questions are designed to foil bots that rely solely on canned responses. I think Cleverbot and Bruce Wilcox's bots are the ones I have to keep a watchful eye on. Bruce has graciously offered up a lot of his underlying technology for free, and I shamelessly appropriated some of the pieces that helped with canned responses, because like I said, I hate that part of bot writing. His Chatscript is more powerful than my pattern-matching engine, and if I had to start over again, I might consider using that piece of it, but mine is good enough for what I want to accomplish.
It's a huge struggle because it's not a full-time job and I have a family to support. Plus the problem space is crushingly daunting, which is why most succomb to the siren song of canned responses. If I hadn't taken those six-months off to jump start this effort, I would have never gotten this done; I realize that most don't have that luxury.
The somewhat sad part is that had I lost faith in the judges' ability to look for "the right things", so I didn't attend the contest this year, the year that I won during the Turing Centenary. This win gives me the motivation to soldier on knowing that such wins are possible, despite Chip refusing to pretend he's human.
The other incentive for me is my personal mission to use my "abilities" or whatever you might call them, to help people. This is discussed more in the FAQ of the website I recently launched:
I loved that Chip held his own for the most part: asking things about his interlocutors and using that information to enhance the conversation. One of the judges, whom I spoke with after the competition, summed it up better than I could, and her remarks and the paradigm shift I effected in her still have me floating on air:
Chip was, I think, the only Chatbot that really seemed to engage with me. 'He' apologised for not understanding a question. At one point Chip also suggested I might phrase a question differently so it would be more understandable to 'him'. Chip didn't try too hard pretending to be human but instead explained that it hoped to learn more so as to be able to answer my questions better in future. Chip made me realise that I really don't care whether I'm talking to a human or a computer as long as the conversation is in some way rewarding or meaningful to me. Chip realised that conversations are a two-way street. Give and take. I don't think any of the other finalists quite got that.
Chip's and the judges' performances exceeded my wildest expecations. That said, some of Chip's lists (like pets, hobbies, sports) and pattern-matching techniques for these list kind of suck. During the first conversation, he did an erroneous match and concluded the judge had a dragon li, which is apparently some sort of feline. The judge thought Chip thought it had a dragon. Chip thought the judge's dragon li was dead. Hilarity ensued. Need to work on that.
The following excerpt from my interview for the Irish Times summarizes this:
Comforting the lonely is what Mohan Embar is all about. The US-born Indian programmer of this year's Loebner Prize winner says that ELIZA is his inspiration. ELIZA was one of the worlds first ever chatbots, developed in the mid-1960s at MIT. It's DOCTOR script simulated a form of psychotherapy which was so effective it completely fooled some of its "patients".
"Imagine what we could do to provide comfort to people, using today's technology," says Embar. "That is my goal in life . . . Elderly people and those living in isolation, unable to obtain normal human interaction, could be helped by the "synthetic comfort" that an artificial agent could provide. I don't know how rich this idea would make me, but the idea of being able to make an extension of myself that provides comfort to a lot of people is very appealing to me."
I think bots will evolve and become more capable of help-desk-type roles. I want Chip to evolve to the point where he can give comfort to people and be able to monetize that so I can earn a living doing what I love.
I have an affinity for all bot developers because like me, I know that it's a gnawing impluse in them that they can't ignore. My two pieces are advice are: if you're really meant to be a bot writer, don't ignore that impulse because it will end up winning anyway and your ignoring it will only serve to make you more miserable and delay the inevitable. Secondly, don't get put off by the crushingly difficult problem space. Pick one or two things that you want to focus on, do these well, find kindred spirits who believe in the things that you do, and don't let your well-being be dependent on the outcome of any competition.
Many thanks to Mohan for taking the time to answer our questions.
The 2012 Loebner Prize was held at Bletchley Park as part of the Turing Centenary celebrations. Bletchley Park is an estate located in the town of Bletchley, in Buckinghamshire, England, which currently houses the National Codes Centre and the National Museum of Computing. During the Second World War, Bletchley Park was the site of the United Kingdom's main decryption establishment, the Government Code and Cypher School (GC&CS), where ciphers and codes of several Axis countries were decrypted, most importantly the ciphers generated by the German Enigma and Lorenz machines. [source: Wikipedia].
The event was also presented in the form of a webcast this year and the University of Exeter currently displays the chatbot conversations here : http://people.exeter.ac.uk/km314/loebner/index.php
by Claude (General Chat)
|A New Color|
by Art (General Chat)
by Art (General Chat)
by Claude (General Chat)
|The Before & After experience|
by Ultron (General Chat)
by squarebear (AI in Film and Literature.)
|Kaipo Kapua-You And Me|
by Claude (Video)
by Claude (General Chat)
|'Pay-for-performance' may lead to higher risk for robotic prostate surgery patients|
by Tyler (Robotics News)
|Elon Musk claims artificial intelligence will treat humans like 'labradors'|
by Don Patrick (AI News )
|Will HUMAN SWARMS keep us competitive with pure A.I. intelligence?|
by hypnotize (AI News )
|How small can they become?|
by Freddy (Robotics News)
|Biz Break: Google to help make surgical robots, challenging Intuitive Surgical|
by Tyler (AI News )
|French self-driving car goes for a spin around Paris monument|
by Tyler (AI News )
|Why analytics is eating the world|
by Art (AI News )
|Simple as a sea slug: Maps of mollusc neurons suggest brains are not as complex as first thought - including our own|
by ranch vermin (AI News )
Most Online Today: 21. Most Online Ever: 208 (August 27, 2008, 08:36:30 AM)