Listen to this article
Artificial intelligence is creeping into our daily lives in unexpected ways. It is not just transforming online services with innovations such as Apple’s Siri voice recognition app, which will send emails when you instruct it to, or Microsoft’s Skype translation services, which enable you to communicate online with people whose languages you do not speak.
Wider applications of artificial intelligence, such as image and pattern recognition (classifying data or objects based on common features), natural language processing (how computers understand and respond to human speech) and machine learning (when software learns something without being programmed to do so) will soon be featuring in many products and services.
The Connected Business examines some of the more unusual uses of artificial intelligence that may be coming to a home or office near you:
In recent years, pest control company Rentokil Initial has been experimenting with rodent traps equipped with sensors and WiFi. These send data to a command centre, which the company has built with partners Google and PA Consulting.
A member of staff is only sent to a trap once the machine has told the command centre it has caught a rat or a mouse. This is more efficient than routine patrols, which would often find empty traps.
Rentokil Initial now has more than 20,000 such devices in 12 countries. It has collected more than 3m pieces of data with these so far. These could be used to finesse the company’s digital pest control services with a dose of artificial intelligence, says Tim Shooter, an independent technology consultant who worked with Rentokil Initial on the pilot project.
By blending information from the traps with weather and mapping data, it might be possible to better identify rodent breeding or migration patterns and identify infestation-risk hotspots before they develop, he says.
“That would mean a significant shift away from reactive pest control services . . . in favour of proactive services that tackle problems before a customer’s even aware of them,” says Mr Shooter.
IntelligentX Brewing claims to have created the world’s first beer to be brewed using artificial intelligence.
The company is a joint venture between creative agency 10x and machine learning specialist Intelligent Layer. The recipes for the company’s four products — Golden AI, Amber AI, Pale AI and Black AI — change over time, based on customer feedback interpreted using a machine learning algorithm.
Codes printed on the bottles direct customers to a Facebook Messenger bot, which asks questions such as: “How would you rate the hoppiness out of 10?” The responses are then interpreted by the algorithm and the findings are passed to the brewers, who tweak their recipes accordingly. The questions change based on the responses the algorithm finds most useful. IntelligentX’s beers can currently be purchased from UBrew in Bermondsey, London, but will soon be available to order online.
While security company Cocoon’s devices use the motion sensors and cameras one might expect, they also detect sounds and vibrations — including low-frequency signals inaudible to humans — and use machine learning to understand the noises that are usual and those that may signify a break-in.
Every home has a unique sound fingerprint, says Cocoon co-founder and head of software John Berthels: this may include lorries rumbling by, the central heating switching itself on and off or a pet moving around. The devices gradually build up a picture of what is “normal” for each house.
If noises deviate from the established patterns — a back door being forced open or a window breaking — the device will send an alert to the users’ smartphone, prompting them to check their home on a live video link, set off a high-pitched alarm or call the police. About 850 people took part in Cocoon’s Indiegogo crowdfunding exercise, raising almost $250,000, and the devices are now on sale online for £299 ($399).
In September 2016, US toymaker Mattel unveiled Hot Wheels AI Intelligent Race System, a twist on its much-loved line of toy racing cars which dates back to 1968.
Unlike slot-car systems like Scalextric, which keep cars on track by means of a pin, Hot Wheels AI has sensors on the underside of each car that interpret a gradient pattern printed on the track.
The AI behind this is no more complex than the technology that guides a robot vacuum cleaner, but it means players can race against self-driving, computer-controlled cars and those controlled by other humans.
The cars are larger than their predecessors and require a games controller. Players can also launch virtual hazards such as oil slicks and tyre blowouts to sabotage their competitors.
The toys are now available on retail sites and Mattel will be hoping for a Christmas hit.
Get alerts on Technology when a new story is published