#DTTT Expert Igor Dutra is an experienced and creative professional with a proven record in user-centered design, innovation, design thinking, lean UX, information architecture, interaction design, visual design, agile methods, web development, and technology. Igor will be joining us on stage at #DTTTGlobal focusing on 'Fostering Innovation through Technology & Digital Trends'.
I’m fascinated with the way my children interact with Amazon’s Alexa during our convoluted dinner time. They are 6 and 8-years old and for them, Alexa is definitely a person who’s there to serve you - as long as she listens! They fight to trigger her attention expecting her to listen to whoever shouts louder. And when one of them is talking to Alexa, the other starts making requests confusing the poor digital lady who cannot identify who’s talking what. When one of them manages to get Alexa to listen, it’s something like “Alexa, play this song…”, “Alexa, tell me a joke” or “Alexa, what’s 10230492 + 1?” – the 6-year old thinks that’s hard math.
What does Alexa do?
I’ll try to explain what Alexa does in the background – more details can be found on Alexa Voice Service (AVS) documentation. And by the way, similar services are provided by IBM Watson, Microsoft Cognitive Services, and others. First, when my kids say “Alexa” – she starts listening, expecting instructions. Alexa then transcribes what they’re saying and identifies what they want to do (an intent). If they say “tell me a joke”, “say something funny” or “make me laugh”, Alexa will perform the action associated with the intent and tell them a random (sometimes funny) joke. You can say “tell me kids jokes” and she will say something appropriate for kids, and sometimes she’ll even pick regional accents! If Alexa cannot identify the intent she’ll apologise and say that she doesn’t understand. But Alexa learns all the time, improving the results with millions of new phrases every day, grouping, training and classifying correct and incorrect intents - yep, that’s machine learning.
Developers can create Alexa apps (called skills), define their own intents e.g. “Alexa change the living room temperature to 22 degrees” using a smart thermostat skill. Skills can also contain decision trees so users can navigate using voice commands. A flight search skill would ask where are you flying from, flying to, date, number of passengers, etc and will read aloud the results – pretty much mimicking the user interface using voice commands and reading instructions aloud.
Voice interfaces are becoming part of our lives the same way I grew with the infant web in the 90s. It’s on all smartphones (think Siri, Google, and Cortana), it’s the ubiquitous Amazon ecosystem, and (Tesla!) cars. As the technology matures, it’s inevitable. But… Are you sure you’d like to hear 10 flight results, with dates, times, connections and prices? If you go to a shop and have a chat with a travel agent you’ll probably have just a couple options tailored to you but you’ll be able to expand and explore alternatives as part of the conversation. For example, you can say you’d like to go to Mexico, but there’s a flight connection in the USA and you mention you’ve never been there - the travel agent can organise a stop-over or even suggest more appropriate flights based on that new requirement. And if you search online you can add filters, open multiple browser tabs, add bookmarks and explore multiple options yourself. Both alternatives would be really tricky to emulate using voice interface today!
Creating the human dialogue is the real challenge. There’s a lot of skepticism of ‘talking’ to a machine (my kids don’t seem to be bothered with that though), and research shows people feel too uncomfortable to talk to a machine, especially in public. The machine’s dialogue would also have to go beyond the decision tree, adapting, being ‘nice’ and sociable, asking the right questions at the right moment and extracting the information to make decisions. Imagine the travel agent from my previous example talking to a couple or group of people about their holidays… The dialogue complexity grows exponentially – and also the moderation, negotiation and judgment skills required.
Humans are naturally very good at communicating (especially women!). We are curious, we try to establish contact and be socially accepted. We understand when we shouldn’t say something inappropriate because it’s part of our moral etiquette. But, can’t computers learn social and empathy rules and simply be more… human?
I think with all progress and investment in Artificial Intelligence and related fields, computers will be able to talk (and listen) like humans in a not too distant future - Alexa is just the beginning. Technology breakthroughs are happening at a very fast pace. Recently Microsoft claimed their speech recognition system is now as good as human and a couple years ago Google open sourced their software library for Machine Intelligence (called Tensorflow) which has direct applications in Natural Language Processing - basically it has a lot to do with the ‘human dialogue’ challenge I mentioned earlier. It’s exciting times because all this technology is available to everyone - it just requires some imagination to solve real people’s problems.
We can start to imagine all sort of innovative services that can transform travellers’ experiences. Planning your trip talking to a virtual travel advisor which knows everything, find the best deals and is always available - and most important acts and talks like a real person. Or at the hotel where you can order mojitos on the beach talking to a smart sun bed. It can also be your virtual friend who explores a city together and makes your trip more exciting and fun.
And I truly believe at some point machines will also be able to read our minds and communicate in some form of telepathy. I’d love to think about a hot relaxing bath while I’m skiing during the day and when I get back to my room the bath tub will be ready for me!
My kids would probably ask… Alexa, what’s coming next?
Computer speech has evolved quite a lot, this is an example of my first contact with a ‘talking’ computer some point in the late 80s
Discussioncomments powered by Disqus
More from #DTTT
In January we present:10 Takeaways on Creating a Killer Content Strategy
When it comes to creating content, there are key factors that every brand should consider. Amongst other elements, one of the most important points about content is the need for a strong strategy.
In December we present:6 reasons why to become a Launchpad member for 2019
Here at the #DTTT we are proud to offer The Launchpad, a wealth of industry knowledge, insight and examples, and it is going to be bigger and better in 2019. Here’s why your team should sign up today!
In November we present:#DTTT Global Takeaways
Wow, this year’s #DTTTGlobal has gone above and beyond our expectations! Such an inspiring two days, packed full with valuable insights and knowledge from some truly dynamic DMOs who joined us from around the world (and of course some authentic Finnish sauna time in between!). It’s so great to see destinations developing and evolving their […]
In November we present:Welcoming visitors back to Puerto Rico
Discover Puerto Rico is the newly established, non-profit destination marketing organisation for the island of Puerto Rico, responsible for all global sales and marketing.
In November we present:5 reasons why we’re excited for #DTTTGlobal 2018
#DTTTGlobal is just over a week away and we are all getting very excited here at DTTT HQ!
In November we present:Marketing Canada’s first national park with a data-driven approach
Diane Bures, Director of Consumer Marketing at Banff and Lake Louise Tourism will be joining us at #DTTTGlobal and we are looking forward to hearing about the different marketing initiatives it has developed to enhance the visitor experience while always encouraging responsible tourism.#data drive strategy #responsible tourism #canada #storytelling