Robbie Allen, the founder and chief executive officer of Automated Insights, is playing Scrooge when it comes to media reports about a supposedly big advance in artificial intelligence.
A group of Russian programmers created quite a stir earlier this week when they were reported as the first to beat the so-called “Turing Test,” convincing humans that the “speaker” was human, not a computer.
But Allen of Durham-based AI, a provider of robotic writing and reporting software, reacted with a “Bah, humbug!” of sorts when asked by WRAL TechWire about the reported breakthrough.
“Our technology passed the Turing test a while back, just not in front of the Royal Society in London,” Allen said.
“A study showed our content was indistinguishable from that of a person. What chatbot achieved is a little dubious in my mind. They convinced one third of the judges that it was a 13-year-old boy.
“A bigger accomplishment would have been to convince 13-year-old boys!”
The Independent in London reported:
“Turing Test breakthrough as super-computer becomes first to convince us it’s human
“Eugene Goostman, a computer programmer pretending to be a young Ukrainian boy, successfully duped enough humans to pass the iconic tes.t”
Exclusive WRAL TechWire reporting on robotic content:
- WRAL TechWire Insider and Automated Insights executive Joe Procopio delivers speech to journalists on robots in their future
- Joe Procopio’s report on robotic journalism as the coming wave.
Turing refers to computer genius Alan Turing whose pioneering efforts helped defeat Germany in World War Two. He devised the so-called Turing Test as a benchmark against which advances in artificial intelligence would be measured.
Could computers really think?
International media picked up on the story, but Allen says the success of “Eugene Goostman” – a chat program tested by a panel of humans at the University of reading – pales in comparison to what his company has achieved.
No Complaints About “Sounding Like a Robot”
Asked if the Goostman breakthrough meant more credibility for automatic intelligence and robotic reporting, writing, Allen was dismissive.
“Our platform, Wordsmith, produced over 300 million personalized stories last year for big companies like Yahoo! and Samsung,” he replied. “I’m not sure a chatbot, which was designed to ‘know anything, but his age also makes it perfectly reasonable that he doesn’t know everything’ advances the cause.”
Allen was referring in the “know anything” comment to a quote from one of the programmers cited in The Independent story.
In terms of significance for the advancement of automated intelligence, Allen said his company has support from some very high-profile customers, such as Yahoo.
“Anytime announcements like this are made (e.g. when news of ‘quakebot’ came out a couple months ago), companies start looking around for solutions that can take those advances into business,” he said.
“We are the only company around that transforms proprietary data into personalized content at scale. And none of our dozens of clients have complained about the content sounding like a robot.”
By the way, as noted by The Independent, there is a contest called “A Long Bet” that focuses on the advancement of artificial intelligence: “By 2029 no computer – or “machine intelligence” – will have passed the Turing Test.” The prize at stake is $20,000 with Lotus founder Mitchell Kapor having made the bet and futurist Ray Kurzweil taking him up on it.
Bettors might want to bet ON Automated Insights, according to Allen.
“Despite some of the skepticism around robot writers,” he said, “I haven’t found anyone willing to bet against what we’ll be able to accomplish in the future with our Wordsmith platform.”