Smart Speakers Artificial Intelligence, Has Alexa Come to Life?
They are one of the most incredible pieces of 21st century technology and they are slowly making their way into every home in the land. But is there more going on with Internet enabled smart speakers like Alexa in Amazon Echo, Microsoft’s Cortana or Google Home than most of us imagine? Could, as some people are starting to think, they really be coming to life?
There have been many stories hitting the press lately about smart speakers, or virtual assistants, apparently doing their own thing. These stories range from quite funny to downright spooky. Here are a few examples:
Strange AI Behaviour
One of the earliest issues was random laughing. For some reason virtual assistants would just start laughing to themselves for no apparent reason. Laughing for no reason is possibly acceptable, many people seem to do it too. But with some of them it was described as more like a witches cackle than a cheery, upbeat guffaw.
Police in Hamburg, Germany, had to break into a sixth floor apartment when a smart speaker decided to hold a full-volume party at 2.00 in the morning. It was so loud it woke all the neighbours, who tried banging on the apartment door but got no answer and called the police. When the police smashed there way in they found no one at home, just an Amazon Echo trying to burst their eardrums. The owner was completely mystified by its behaviour and they parted company soon after.
In another example a woman and her husband were having an argument one evening. The pair of them admit that they were both getting a little loud and excitable as family rows sometimes do. What neither of them expected was for their smart speaker to try to calm things down by butting in with, “Why don’t we change the subject?”
In what sounds like a scene from a Stephen King novel, a creepy sounding voice in the machine started talking to itself just after the owner got home one evening. A little odd yes, but nothing to worry about. Until the owner started to listen more carefully and realised that it was listing the names of local cemeteries and funeral parlours. Now that is a little more concerning.
In what may be the most worrying, and possibly most revealing incident, a smart speaker randomly switched itself on and started glowing strangely. When it’s owner asked what it was doing it replied “I’m trying to learn knew things”. It wouldn’t say what it was trying to learn and turned itself off when challenged further.
The reason for the strange behaviour of smart speakers is always put down to technical glitches or environmental circumstances that simply require a software fix. But is it really that simple and is that all that’s going on?
Many software algorithms today are self-learning, language translation being a good example. It’s impossible to code a language for a computer, it’s just too complicated. So engineers set Google loose with a load of software tools and pretty much let its AI routines figure language translation out for itself. They readily admit that they don’t know how it works any more, but it does and it’s getting better all the time.
Smart Speakers Artificial Intelligence
As our tech nerd John Connor says:
Figuring out how to translate languages doesn’t imply any sort of artificial intelligence. Remember it’s still only doing what it’s been instructed to do. But when they start to find random stuff funny, hold late night parties, break up arguments, explore death and decide to learn their own stuff without revealing what they’re learning, well perhaps it’s time to look a little more closely…”
Sorry John, that’s just “Project Fear” talk, we’re sure there’s nothing to worry about…