Chatbots can already read you the news, order you a takeaway, or hail you a cab. Now they're heading for your kitchen, with the launch of Jamie Oliver Skill, a virtual sous-chef developed by digital agency AKQA and the Jamie Oliver Group for the Amazon Echo.
The Echo is a voice-controlled wireless speaker that is Amazon’s answer to Apple’s Siri, Google’s Now and Microsoft’s Cortana. Microphones on the device transmit your voice then a virtual assistant called Alexa—now equipped with a British accent for the UK market—responds automatically using artificial intelligence tech to create a simulated conversation.
Jamie Oliver Skill is one of the early additions to Alexa’s array of capabilities. More than 160 recipes are divided into eight main ingredient categories: vegetarian, pasta, eggs, fish or meat (beef, chicken, lamb and pork).
When users ask for a meal recommendation, Alexa responds with a selection of suggestions. It provides further details on the difficulty and timing of the recipe when prompted.
Once a dish is selected, an email is sent to the user with a summary of the recipe, a full list of ingredients and a link to jamieoliver.com for further instructions.
How Natural Language Processing works
“Speech recombination is what gets the input into the services,” says Andy Hood, head of emerging technologies at AKQA. “Actually understanding that speech is the crucial bit.”
The software analyses human language through Natural Language Processing (NLP), a programming system that provides automated responses by spotting patterns in speech.
“Natural language processing means that an AI is taught to interpret the content of phrases and sentences,” says Hood. “It does that through training, rather than just being told this is what these things mean.
“Rather than telling something how to solve a problem, you tell it that you have the problem and then it learns how to solve it itself.”
Alexa adapts better to your individual speech through each interaction thanks to machine learning. The more data it has to analyse, the greater the variety of tailored answers it has reserved.
“Natural language processing that changes input into conversation,” is how Hood describes it. “That's at the core of it. Once you've got the input and you've understood what the request is, then it's reasonably standard bot behaviour after that.”
The Echo: what's next?
More recipes and further filters covering timing and dietary requirements will be added to Jamie Oliver Skill later this year, but the range of responses will remain restricted.
“The functionality of the Echo is fairly limited at this point,” Hood admits. “Which is fine, because it's day zero for this kind of thing. These things don't emerge fully formed.”
The addition of a message from Jamie Oliver at the end of the conversation and the subsequent delivery of an email message were attempts to add authenticity to the interaction and overcome the current limitations of the technology.
“We’re trying to make the conversation feel more human and add some multiple layers to how the interaction feels."
Voice command has already evolved beyond the early novelty value of Siri to become acceptable mainstream behaviour. Hood has high hopes for the technology and envisions a future connected kitchen full of smart cooking appliances.
“The interesting bit happens when it stops being absolutely brilliant, and almost blends into the background; that's when you really see full value from things,” he says.
“That's when people stop doing things just so they can say they did them, and they actually have to produce value with them."
Voices becoming brands
In his keynote speech at Google I/O this year, Google CEO Sundar Pichai revealed that one in every five searches made in the US with the company’s Android app was now a voice query, and that this share is growing.
Hood believes that brands are only scratching the surface of the technology’s commercial potential.
“This is a chance for a brand to be in someone's living room”, he says. “The device is almost irrelevant,” he says.
Chatbots have already been heralded as the new apps, a sentiment that Hood seems to share.
"When iPhone opened the app store all those years ago, people wanted to be on the device. That was the client's desires.
“The thing with voice is, you could hide the device. It could be behind someone glasses or behind a book or something. it doesn't really matter about the device itself, the key is the voice.”
That ubiquitous voice presence in living rooms or kitchens changes the relationship with consumers into a conversation that is emotionally direct.
“The voice isn't merely the means of input,” says Hood. “It's actually the user experience, and therefore the brand.”