No, we are (probably) not about to be replaced by robots
Are the robots about to replace us all, both in physical and mental work? If you read the press you are likely to feel a sense of imminent doom. But Matthew Taylor, Chief Executive of the Royal Society of the Arts, makes a good point:
“Technology won’t happen as quickly as people say. Its easier to get a book publicised if it says AI will replace us by next Tuesday than if it says its going to take a long time – that voice isn’t getting heard.”
I understand the Kurzweil argument on exponential growth. If Moore’s Law still holds, stating computing power doubles every two years, then in just 20 years computers will be one thousand times as powerful as they are now.
Given that computers have already beaten humans at Chess, Jeopardy and Go then, once they are one thousand times more powerful, they will surely be at least as intelligent as humans?
However it is now 21 years since Deep Blue beat Gary Kasparov, then the world champion, at chess. That means computers are now one thousand times more powerful than at that pivotal moment, and they still are not able to pass the Turing Test – to be indistinguishable in conversation from a human.
Last September I attended a talk entitled “will your next work colleague be a chatbot”, suggesting that your company was likely to use a computer to perform the kind of tasks you do. By chance, the next day I found myself engaging with a chatbot on my bank’s website.
It was infuriating. The software seemed unable to do anything beyond referring me to pages I had already read, and clearly did not understand what I was trying to ask it. It was the most frustrating experience I’d had in 25 years with the bank First Direct (normally so customer friendly and a joy to talk to). I was pleasantly surprised, next time I used the site, to find real people were back, who could quickly solve my problem.
The term chatbot is of course ridiculous. They may be pretty good at providing information but the one thing that chatbots are totally useless at is chatting.
As management guru Tom Peters puts it in his latest book The Excellence Dividend, there is nothing like “the quality of fully engaged employees providing personalized service that makes you smile as it is delivered and creates fond memories that last.”
He continues “excellence that translates into an emotional bond with customers and communities in a way that cannot—and I predict will not—be replaceable by algorithms in the foreseeable future.”
Tom is not a luddite. I remember hearing him in 2000 predicting that back-office white collar jobs were about to be outsourced or replaced by technology. He was right about that and I think he is right about the advantages that people still have.
I have visited two best-practice manufacturing plants in the last couple of years and have been intrigued to find neither were dominated by robots. At Toyota they were actually using less robots than a few years ago. “They aren’t as flexible as people”, they explained.
At Buchler, who make amazing cutting-edge grain-sifting technology, they were instead exploring “cobots” – Collaborative Robots which support people to do a better job rather than replacing them.
Indeed back in the world of chess a very interesting event was held in 2005, which included computer programs, grandmasters and combinations of people and machines. The tournament was won by two chess amateurs, taking part with three different software programs. Although computers were arguably 16 times more powerful than when Deep Blue beat Kasparov, a combination of people and software beat both the grandmasters and the computers working on their own.
And that is perhaps a more likely future, not AI but IA – Intelligent Augmentation, humans working with computer software. There is a fascinating article here on how AIs are best at finding answers. Humans are best at finding questions.
Take Google. We all use it to help us find the web sites we are looking for. But when was the last time you clicked the “I’m feeling lucky button”, which takes you to Google’s top choice? Even after 20 years, and billions of dollars of investment in the algorithm, we don’t trust Google to provide the answer – instead we use it as a support to give us options to choose from. (“I’m feeling lucky” is one of only two buttons but is apparently chosen less than 1% of the time.)
I read of one piece of software that analyses responses from the person you are talking to and advises on whether you talking too much, or asking enough questions. In call centres its been shown to improve performance by over 20%. It is a great support but nobody is suggesting, for all its expert analysis, that the software should carry out those conversations itself.
So do I expect chatbots to have meaningful conversations with customers anytime soon? Not in the next decade in my opinion.
But will there be some fascinating software to support your people to enhance their abilities and provide a better service? Yes, that makes sense and is actually quite exciting.
About the author – henry Stewart: