From cell phones to espresso machines, smart devices are ubiquitous in modern day living. And tech companies are racing to offer the slickest, most efficient digital assistants for everything, from managing your calendar to monitoring traffic. Richard Waters of the financial times explores some of the many artificial intelligence companions on the market. Are they really working?
From cell phones to espresso machines, smart devices are ubiquitous in modern-day living.
And tech companies are racing to offer the slickest, most efficient digital assistants for everything from managing your calendar to monitoring traffic.
Richard Waters of the explores some of the many artificial intelligence companions on the market.
Are they really working?
Let's take a look.
Helpful or threatening?
Cute or scary?
Whether we like it or not, these machines are becoming more commonplace and fundamental to the way we live our lives.
The very main feature of this robot is the fact that it is entirely controlled by the human voice.
Are you married?
Talking to computers has been a dream of technologists and storytellers for decades.
Even if it can sometimes seem a bit scary.
Encounters with disembodied intelligences are no longer confined to science fiction.
Intelligent agents, digital assistants, chatbots -- there's a growing array of these smart companions that are eager to become our guides to the digital world.
Google is one of the powerful tech companies who believe the new technology will be as significant as the first phase of the Web.
It's the wild, wild West.
Everyone's trying to figure out, like, what is the right morality?
People really got very, very familiar with, you know, accessing information on the Web and, you know, using browser to kind of do things that they never could do, you know, before that.
Um, so I see that kind of a step-function change in how people will actually react and interact with computing.
But I also see a lot of experimentation needed to get to next steps here.
I'm here in the studio at ToyTalk.
And this is where the voices of chatbots are recorded.
Humans record thousands of lines of script.
And when you talk to an artificial intelligence in your smartphone or some physical object, uh, the machine will try to anticipate and understand what your interests are and then feed lines back to you that it thinks are in context.
Um, so I have the first physical embodiment of that here.
Um, this Barbie doll actually has a chatbot implanted inside.
[ Ding ]
Can you say that again?
Do you want to play a game or talk some more?
I really like clothes.
[ Ding ]
What's the point if it's not fun?
[ Ding ] Why did the belt go to jail?
[ Ding ]
I don't know.
Why did the belt go to jail?
[ Ding ]
Because he held up a pair of pants.
[ Laughs ] Want to hear another one? Uh-oh.
I can't find a Wi-Fi network.
It's not quite Hal.
But chatbots like Barbie could be a forerunner of how we will all, one day, be talking to computers.
These programs are sometimes given a veneer of human personality and are designed to engage you in conversation, usually by feeding back canned responses.
They fit easily into today's widely used messaging systems, where they can chat back and forth with humans in text form.
And then one of the interesting qualities of this trend is that it actually largely started, um, in the East and has come here to the West.
Um, what WeChat in China in particular, um, is the portal to the Internet in mainland China.
And the number of bots and businesses that exist in texting there has been an enormous number for now half a decade.
Just this year, in 2016, we're gonna see that happen on the major platforms here in the States.
Bots often come with personalities already baked in.
They can appear in mobile chat apps.
Or like these, they can be designed as characters in video games, creating a new form of interactive storytelling.
Facebook and Microsoft have both put bots at the center of their plans recently, seeing the beginning of an entirely new way of living with computers.
All the big tech companies are racing to experiment with different interfaces.
There's the Amazon Echo, a black cylinder that sits on your kitchen table and orders groceries; Apple's Siri, a question-and-answer system that was developed for smartphones; Google Now, a predictive assistant that tries to anticipate what you'll want to do next and give you the information before you even ask; Facebook M, a text-messaging system which relies on people at the moment, but Facebook wants to replace with artificial intelligence; and Microsoft Cortana, a personal assistant that is trying to make the move from the PC to the mobile world.
What's the weather gonna be like tomorrow?
The bar for success for all these experiments is set much higher than for many of today's technologies.
To give you an example, if, you know, you search for something, and you got an article, you got a pointer to a webpage that's not as relevant, it's one thing.
But let's say the assistant said, 'Hey, you know, it's time to leave for your flight.
And guess what.
Your flight's delayed.
No problem,' and then it turns out it's not delayed and you miss the flight, that's a problem.
The range of experimentation suggests the field is still wide open, with upstarts eager to challenge the big tech players.
Voice-controlled digital assistants like Hound represent the first generation of these intelligent assistants.
Chief executive Keyvan Mohajer has been working on this dream for 10 years.
When we designed Hound, it was very important to us that the interaction between the user and our application is very natural.
So when you talk to Hound, we want the response to be natural as if a human being is responding to a question.
We wanted the responses to be grammatically correct, not too long, not too short, and delightful.
Uh, okay, Hound.
[ Ding ] Show me coffee shops that are within a half a mile of where I am and that are open now.
[ Ding ]
Here are several coffee shops within 0.5 miles.
I'd like one that has the best coffee and where I can sit outside with free Wi-Fi.
You didn't say, 'Okay, Hound,' so it didn't... [ Ding ]
Companies like Hound believe voice control will become even more important with the much-predicted Internet of Things...
Make me an espresso.
...when many everyday objects become smart and connected.
Please place your cup under the spout and say 'ready.'
You can ask your espresso machine to give you, a double-shot espresso.
While it's making the double-shot espresso, you can say, 'How's the weather today?
Who won the game last night?'
So one device can enable its own functionality.
But it can also enable multiple other domains that the creator of the device can think is useful for the end user.
Communicating through voice or text messaging is only part of what makes these new digital assistants and agents intelligent.
Behind the scenes, companies like Google are applying new types of machine learning and analyzing data about you to come up with the best answers.
Despite the considerable hype, the intelligent agents are not quite ready for prime time.
Their language abilities are not perfect.
They don't always know exactly what we want.
But they're making rapid progress.
And this leaves plenty of questions.
What are we giving up in return for the convenience of talking to the machines?
They will be making assumptions about what we should know and when.
They'll be in the background, listening, perhaps watching what we're doing.
They will assume more control over our lives.
But the payoff in terms of a more useful and practical coexistence with the computers could be significant.