Talking machines? No thanks.
Artificial intelligence isn’t as intelligent as you may think, says Hong Kong’s humorist
I first realized that artificial intelligence (AI) didn’t really exist when I started looking at the ads that showed up on my friends’ computers.
You know what I mean. They do a web-search for “hepatitis symptoms” and get an ad saying “20 percent off hepatitis symptoms now.”
They also get ads for things they’ve just bought. “You recently purchased a hot tub: here are other hot tubs to buy.” In Hong Kong that might make sense, I suppose. If you are rich enough to have your own hot tub, you’re probably rich enough to buy them in bulk. “Just leave them in a pile over there.”
But it’s when you get error messages that you realize how dumb computers are: “Keyboard not detected. Press any key to continue.”
This trail of thought was triggered by the arrival in my inbox of a press release introducing a service called AI Accountant – one of more than a dozen features I’ve been sent recently on the automation of accounting.
This worries me. In all my serious in-depth research of the subject (watching TV) I note that all robot films have the same plot. The automatons become conscious and then create problems for humans.
Worse still, people don’t seem to have noticed that the latest tech fad, the Internet of Things (IoT), will allow computerized objects to communicate with each other.
Imagine the scene.
Robot accountant: “I am adding a capital outgoings charge for the new fridge and coffee machine the client bought for the staff canteen.”
Message from fridge: “He’s lying! I’ve been here for years, check my records.”
Message from coffee machine: “Actually he did buy a new coffee machine, but took it to the secret flat he keeps for his mistress. Check the location data.”
Message from car: “True. I’m parked outside her flat now, ha ha.”
In ethical terms, this will be good. Thanks to the IoT, the robot accountant will be able to do a more comprehensive audit than the client ever thought possible.
Humans will hate it.
The only answer is for robot accountants to be tightly controlled with strict boundaries set for every activity.
On the downside, I tried that when I became a parent.
Plan: “I’ll teach my kids to be polite, multilingual critical thinkers who will save the world.”
Reality: “Phew, I managed to stop the kids running naked out of the apartment today.”
By chance, this columnist was sent to cover a robotics conference at Hong Kong Science Park and decided to research the subject further. I found two schools of thought.
One was represented by a man who talked breathlessly about being on the verge of achieving machine consciousness, but he did not fill me with confidence as he had failed to do up his jacket buttons correctly.
The alternative view was represented by a man who said that there was no such thing as AI. It was just a term for a series of dumb, automated decisions which gives the impression that human thinking has taken place.
He pointed out that the father of modern computing, Alan Turing, referred to the test of the AI concept as “the imitation game” – it’s not thinking, but pretending to think.
That reminds me of an important principle that I taught my daughter, who recently started her first paid job.
Never sign a contract without pretending to read it first.
She looked puzzled.
Life’s too short, I said, and showed her a 29-page book contract I signed the previous day. Long, boring and unreadable.
A bit like my book, probably.
Nury Vittachi is a bestselling author, columnist, lecturer and TV host. He wrote three storybooks for the Institute, May Moon and the Secrets of the CPAs, May Moon Rescues the World Economy and May Moon’s Book of Choices