Facebook is now using artificial intelligence to automatically generate captions for photos in the News Feed of people who can’t see them.
The tool is called Automatic Alternative Text, and it dovetails with text-to-speech engines that allow blind people to use Facebook in other ways. Using deep neural networks, the system can identify particular objects in a photo, from cars and boats to ice cream and pizza. It can pick out particular characteristics of the people in the photo, including smiles and beards and eyeglasses. And it can analyze a photo in a more general sense, determining that a photo depicts sun or ocean waves or snow. The text-to speech engine will then “read” these things aloud.
A Facebook employee named Matt King showed me a prototype of the service last fall. King, 49, is blind, and though he acknowledged that the service was far from perfect, he said it was a notable improvement over the the status quo. He wasn’t wrong. King showed the system a photo of a friend and his bike as he traveled through Europe. Facebook’s AI described the scene as outdoors. It said the photo included grass and trees and clouds, and that the scene was near water. If the photo had turned up in his News Feed in the past, King would have known only that his friend had posted a photo.
“My dream is that it would also tell me that it includes Christoph with his bike,” King told me. “But from my perspective as a blind user, going from essentially zero percent satisfaction from a photo to somewhere in the neighborhood of half … is a huge jump.”
As King told me, the system doesn’t always get things right. And it hasn’t yet reached a point where it generates captions in full and complete sentences. But this will come. Others have already used deep neural nets to do much the same thing. As King pointed out, a service that only gets part of the way there is still important now—Facebook says that more than 50,000 people already are using the service with text-to-speech engines.
Ever wished you could order Taco Bell for everyone in the office using a simple chat bot? No? Well, you’ll soon be able to it anyway.
Taco Bell has blessed the world with TacoBot, a chatbot for the popular workplace chat app Slack. Tell TacoBot what you want, and it keeps a running tally of your order, just like that screen at the drive-thru. When you’re done, pay through TacoBot and pick up your order at the Nearest Participating Taco Bell.
If that sounds like it would be handy for your team—or, you know, your “team”—you can join the waiting list. Be warned, though—it might be a few months before you revel in the miracle of this advance in productivity tech, says Martin Legowiecki. He’s the creative technology director at Deutsch, the agency that built TacoBot. And even after you’re in, you’ll still have to send someone with the Taco Bell app on their phone to fetch the order. So, much like the Tacocopter before it, the real bot-ified future of meat, cheese, and tortilla still hasn’t arrived.
But fear not: Legowiecki says delivery is something Taco Bell is working on, too.
Really Slacking
TacoBot is one of the most gluttonous manifestations yet of the bot-powered future of work that many companies predict. In China, people already use the popular instant message app WeChat for all sorts of daily tasks, from checking their bank balances to ordering cabs to buying sneakers. Silicon Valley entrepreneurs are betting that chatbots will replace apps here in the US, too.Slack, with competitors like Hipchat, hope this chat revolution won’t just change consumer behavior but the way work gets done at corporations with lots of money to spend on office supplies, company lunches, and whatever else big companies spend their budgets on. Slack is taking this so seriously that last year it said that it would invest $80 million in companies that build Slack bots.
That vision of the future aside, TacoBot came about almost by accident. Deutsch was contracted by Taco Bell to redesign the company’s website, Legowiecki says. During the course of that project, the Taco Bell and Deutsch teams worked together using Slack. “They fell in love with it, we fell in love with it,” Legowiecki says. “So we though wouldn’t it be great to make Taco Bell available through Slack?”
The Deutsch team built TacoBot using Wit.ai, an online service for building software that’s able understand and respond to human language. (Facebook bought Wit.ai last year to help build its virtual assistant M.) Legowiecki says this means it will be able to bring TacoBot to other apps in the future, such as Hipchat, Facebook Messenger, Amazon Echo, even Apple TV.
In other words, food bots are coming, whether you want them or not. As if you need another excuse to hit the drive-thru.
First reported by The Wall Street Journal, the deal is notable because, well, it’s Apple, the world’s most valuable company and one of the most powerful tech giants. It’s unclear how Apple intends to use the company, but as Moore indicates, the tech built by Emotient is part of much larger trend across the industry. Using what are called deep neural networks—vast networks of hardware and software that approximate the web of neurons in the human brain—companies like Google and Facebook are working on similar face recognition technology and have already rolled it into their online services.
'There are huge implications in terms of making dialogue with computers much more meaningful.'
“We have very real data points showing computers doing a better job
than humans in accessing emotional states,” Moore said in December.
“There are huge implications in terms of making dialogue with computers
much more meaningful.”Moore points our that such technology can be used for everything from security to accessing mental heath. As the Journal explains, Emotient sold its tech to advertisers, letting them analyze how consumers responded to their ads. According to the startup, doctors have also used the technology to determine patient pain, and retailers have used it to track how shoppers react to products in stores.
With deep neural nets, machines can learn to do certain tasks by analyzing large amounts of data. If you feed enough photos of someone smiling into a neural net, for instance, it can learn to understand when someone is happy. And these techniques can be applied to more than just images. They have also proved successful with speech recognition and, to a certain extent, natural language understanding.
Google and Facebook and Microsoft are at the forefront of this deep learning movement. But Apple been pushing in the same direction. In the fall, Apple acquired a startup called VocalIQ, which uses deep neural nets for speech recognition. You might not be able to hide your true feelings from Siri for much longer.
Overseen by ex-Googler Ilya Sutskever and Greg Brockman, the former CTO of high-profile payments startup Stripe, OpenAI has the talent to compete with the industry’s top artificial intelligence outfits, including Google and Facebook—but the company has been setup as a non-profit. “Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return,” Brockman said in a blog post.
The apparent aim is to build systems based on deep learning, a form of artificial intelligence that has proven extremely adept in recent years at identifying images, recognizing spoken words, translating from one language to another, and, to a certain extent, understanding the natural way that we humans talk. Sutskever is a protege of Geoff Hinton, one of the founding fathers of the deep learning movement, who now works for Google.
Deep learning relies on what are called neural networks, vast networks of machines that approximate the networks of neurons in the human brain. Feed enough photos of a cat into a neural net, and it can learn to identify a cat. Feed enough dialogue into a neural net, and it can learn to carry on a pretty good, if sometimes dodgy, conversation. The hope is that top researchers can take this much further. Some believe the method can even be used to mimic human common sense.
Currently, companies like Google and Facebook and Microsoft sit at the forefront of this movement. But OpenAI aims push the state of the art forward without worrying about financial gain. Instead, they intend to open source their work, freely sharing it with the world at large. Recently, Google open sourced the core software engine, TensorFlow, that drives its deep learning services, and just this week, Facebook open sourced its deep learning hardware. They too are looking to advance the technology through widespread collaboration. But you have to wonder if they made these moves because they knew OpenAI was on the way.
OpenAI certainly has the pedigree to make some serious headway in the field. Musk and Thiel are co-chairs of the company. Other backers include Alan Kay, one of the founding fathers of the PC, and Yoshua Bengio, another top deep learning researcher. Altogether, OpenAI says, its backers have committed $1 billion to the project. Sometimes you have to spend money, even if you don’t plan on making any.
Today Howard’s company, Enlitic, said it was going to start working with Capitol Health Limited, a radiology clinic with locations across Australia, to have its software look at X-rays.
Enlitic won’t replace radiologists. Instead, the software is designed to help them do their jobs more quickly and make fewer mistakes. First, it checks each file submitted to make sure the image matches what the technicians say it’s supposed to be—for example, it makes sure that if an image is tagged as a left knee that it’s not actually a right knee. Then, it looks for anomalies in the image.
Depending on what it finds, it assigns a priority to the X-ray and routes it to a radiologist. For example, if it finds nodules on an image of a pair of lungs, it will assign it a high-priority status and route it to a pulmonary radiologist. If it detects what appears to be an aneurysm, it will route the X-ray to a cardiovascular radiologist instead. If it finds no anomalies, it will mark it as low priority.
After a radiologist has reviewed the image, the software will help write the paperwork by auto-generating text for some of the more tedious parts of a report.
Enlitic’s X-ray algorithms are just the latest example of deep learning being put to practical use. Facebook is now using deep learning techniques to caption photos for the blind. Yelp recently explained how it’s using deep learning to optimize what photos it shows on restaurant listings. And similar techniques are also part of what powers Microsoft’s Skype Translate.
And Enlitic isn’t the only company trying to apply artificial intelligence to medicine. IBM’s Watson has been used for research at Memorial Sloan-Kettering Cancer Center and more recently to provide diet and exercise advice. The app Bright.md aims to help physicians speed up routine appointments. But Enlitic’s deal appears to be one of the biggest real-world tests of deep learning’s ability to aid medical diagnostics.
Eventually, Howard hopes the technology will help expand access to medical diagnostics as Capital Health begins opening clinics in Asia. He cites World Economic Forum report that forecasts a severe shortage of medical experts in the developing world if training programs aren’t accelerated. With any luck, artificial intelligence can shoulder some of that burden.
Through acquisitions of companies like Indisys, Xtremeinsights, and perhaps most importantly, fellow chip maker Altera (a $16.7 billion deal), Intel has devoted much its artificial intelligence efforts on baking AI into its into chips, as well as software that powers its 3-D video cameras.
Today, Intel has added yet more AI to its portfolio with the purchase of Saffron Technology. Like many other AI startups, Saffron attempts to extract useful information from huge datasets via algorithms inspired in part on the way the human brain works. But instead of focusing on deep learning, the trendy branch of AI in which Google and Facebook are heavily investing, Saffron is focused its own technique called associative memory. The company was founded in 1999 by former IBM Knowledge Management and Intelligent Agent Center chief scientist Manuel Aparicio and led by former PeopleSoft executive Gayle Sheppard. It has deep roots in the enterprise software industry and cut its teeth selling software to the Department of Defense, such as a system for predicting the likely location of roadside bombs in Iraq.
“We see an opportunity to apply cognitive computing not only to high-powered servers crunching enterprise data, but also to new consumer devices that need to see, sense, and interpret complex information in real time,” Intel New Technology Group Senior Vice President Josh Walden says. “Big data can happen on small devices, as long as they’re smart enough and connected.”
Source: Wired
0 comments:
Post a Comment