Is the Universal Translator Right Around the Corner?

star trek comm badgeWe all love a race. There is something about seeing somebody strive to win that gets our blood stirring. But there is one big race going on now that it’s likely you never heard of, which is the race to develop deep learning.

Deep learning is a specialized field of Artificial Intelligence research that looks to teach computers to learn by structuring them to mimic the neurons in the neocortex, that portion of our brain that does all of the thinking. The field has been around for decades, with limited success, and has needed faster computers to make any real headway.

The race is between a few firms that are working to be the best in the field. Microsoft and Google have gone back and forth with public announcements of breakthroughs, while other companies like Facebook and China’s Baidu are keeping their results quieter. It’s definitely a race, because breakthroughs are always compared to the other competitors.

The current public race deals with pattern recognition. The various teams are trying to get a computer to identify various objects in a defined data set of millions of pictures. In September Google announced that it had the best results on this test and just this month Microsoft said their computers beat not only Google, but did better than what people can do on the test.

All of the companies involved readily admit that their results are still far below what a human can do naturally in the real world, but they have made huge strides. One of the best known demonstrations was done last summer by Google who had their computer look at over 10 million YouTube videos and asked it to identify cats. Their computer did twice as good as any previous test, which was particularly impressive since the Google team had not pre-defined what a cat was to the computer ahead of time.

There are some deep learning techniques in IBM’s Watson computer that beat the best champs in Jeopardy. Watson is currently being groomed to help doctors make diagnoses, particularly in the third world where there is a huge lack of doctors. IBM has also started selling time on the machine to anybody and there is no telling all of the ways it is now being used.

Probably the most interesting current research is in teaching computers to learn on their own. This is done today by enabling multiple levels of ‘neurons’. The first layer learns the basic concept, like recognizing somebody speaking the letter S. Several first-layer inputs are fed to the second layer of neurons which can then recognize more complex patterns. This process is repeated until the computer is able to recognize complex sounds.

The computers being used for this research are already getting impressive. The Google computer that did well learning to recognize cats had a billion connections. This computer was 70% better at recognizing objects than any prior computer. For now, the breakthroughs in the field are being accomplished by applying brute computing force and the cat-test computer used over 16,000 computer processors, something that only a company like Google or Microsoft has available. .

Computer scientists all agree that we are probably still a few decades away from a time when computers can actually learn and think on their own. We need a few more turns of Moore’s Law for the speed of computers to increase and the size of the processors to decrease. But that does not mean that there are not a lot of current real life applications that can benefit from the current generation of deep learning computers.

There are real-world benefits of the research today. For instance, Google has used this research to improve the speech recognition in Android smartphones. But what is even more exciting is where this research is headed for the future. Sergey Brin says that his ultimate goal is to build a benign version of HAL from 2001: A Space Odyssey. It’s likely to take multiple approaches in addition to deep learning to get to such a computer.

But long before a HAL-like computer we could have some very useful real-world applications from deep learning. For instance, computers could monitor complex machines like electric generators and predict problems before they occur. They could be used to monitor traffic patterns to change traffic lights in real time to eliminate traffic jams. They could be used to enable self-driving cars. They could produce a universal translator that will let people with different languages converse in real-time. In fact, in October 2014, Microsoft researcher Rick Rashid gave a lecture in China. The deep learning computer transcribed his spoken lecture into written text with a 7% error rate. It then translated it into Chinese and spoke to the crowd while simulating his voice. It seems like with deep learning we are not far away from having that universal translator promised to us by science fiction.

Mr. Watson . . . . come here.

Watson-supercomputer-635This week IBM cut the ribbon on a “Watson Client Experience Center” in New York City, where along with five other centers it will provide access to the Watson supercomputer. A few weeks ago IBM also announced the availability of what it calls Bluemix, a suite of several cognitive-based cloud services. Several of the articles I read about this announcement say that Watson is bringing artificial intelligence to the world. But it’s not. Watson is a pretty amazing computer system and can do a lot of great things, but the computer is still no smarter than your toaster. You may ask how I can say that since Watson was able to soundly beat the two best Jeopardy champs a few years ago.

Let’s look at how Watson works. First, Watson is a supercomputer, meaning that it has massive computational power and a fast input / output. Watson is configured as cluster of ninety IBM Power 750 servers each of which uses a 3.5 GHz POWER7 eight core processor, with four threads per core. In total, the system has 2,880 POWER7 processor cores and has 16 terabytes of RAM. Watson has a natural language interface meaning that it is designed to be queried by conversation, in the same manner as Apple’s Siri.

Watson uses a hypothesis generator. What this means is that when it is asked something, Watson searches its databases and compiles all of the answers that seem to answer the question posed to it. Through its sheer blazing computational speed Watson can search this entire database quickly. It then ranks the results according to the frequency that it encounters answers. For the Jeopardy challenge Watson was fed with multiple reference sources like encyclopedias, textbooks and all of Wikipedia.

Finally, Watson uses what IBM calls dynamic learning. This means that when Watson makes a mistake, which has to be often when working in something as imprecise as English, Watson can take feedback from the user when told that its answer is wrong. It stores this feedback and uses its ‘learning’ to influence the rankings when it next encounters the same question.

But under it all Watson is no smarter than your desktop computer because there is no actual intelligence in the system, artificial or otherwise. What Watson does to simulate intelligence is to present a friendly language interface and fast computational power to come up with answers to questions. But Watson is only as ‘smart’ as the databases underneath of it. For Jeopardy they did not allow Watson access to the Internet because the internet is full of incorrect facts. Watson has no way of distinguishing between what is true or not true, other than through feedback from users who correct its mistakes. But Watson would be like many of us and would fall for every Internet hoax that hits the web. For example, there was an Internet hoax earlier this year that said that Flo from the insurance commercials was killed, and if Watson was connected to the web it would believe such an untrue rumor based upon the sheer volume of claims made about the hoax.

This is not to say that Watson can’t do amazing things. Imagine Watson paired with Siri. Let’s face it, Siri is okay with driving directions but can quickly get flustered on almost anything else. With Watson’s database behind Siri it would become much more useful in a hurry. And even for driving directions Watson would help Siri be better. Siri is great at getting you between towns, but I’ve noticed that in crowded urban environments that Siri regularly wants you to pull into the wrong parking lot or driveway, and over time Watson would help Siri learn these little nuances of the map through user feedback.

Expect over the next few years to see a flood of new apps that do a better job of working through spoken interface. Already there are interesting new ventures that plan on incorporating Watson. For example, the founder of Travelocity wants to roll out a service called WayBlazer that will help you figure out things to do when you travel. The goal is to help you find activities that interest you rather than being steered to the normal tourist traps. A start-up called LifeLearn wants to build a tool to help veterinarians diagnose pet ailments better. A company called SparkCognition wants to offer a service to help security people spot security risks by having Watson ‘think like a security expert’. Expect all sorts of new programs and apps that take advantage of Watson’s language interface and the ability to quickly search databases.

This is a big breakthrough in that this is the first time that mass computational power will be brought into our daily lives through apps. Those apps are going to start doing things that we have always wanted computers to do. But let’s not forget how quickly computers are getting better. I reported last month on a company that expected to have a desktop supercomputer by 2017 that will be several magnitudes faster than Watson. Within a decade there will be computers everywhere with the power that Watson has today. And let’s also not forget that Watson is not smart and that there is zero cognition in the system. Watson doesn’t think, but rather just searches and compiles large databases quickly. That is incredibly useful and I will be glad to use Watson-based services – but this is not yet anything close to artificial intelligence.

Computerizing our Jobs

Rowa_RoboterI often write about new technology such as cognitive software like Siri or driverless cars. These types of innovations have the potential to make our lives easier, but there are going to be significant societal consequences to some of these innovations. Late last year Carl Benedikt Frey and Michael A. Osborne published a paper that predicts that about 47% of all current American jobs are at risk of being replaced by some form of automated computerized technology.

We have already been seeing this for many years. For example, in the telecom industry there used to be gigantic operator centers with rooms full of operators who helped people place calls. Those centers and those jobs have largely been eliminated through automation. But not all of the jobs that have been eliminated are so obvious. For example, modern accounting software like QuickBooks for small business and more complex software for larger businesses have displaced many accountants. Where a large company might have once had large rooms of accounts payable and accounts receivable personnel, these software systems have eliminated a significant portion of those staffs. And many small businesses perform their accounting functions today without an accountant.

Computerization has also wiped out entire industries and one can only imagine the numbers of jobs that were lost when iTunes largely replaced the music industry or NetFlix and Hulu have replaced video rental stores.

Automation has created some new jobs. For instance, looking at this video of an Amazon fulfillment center we can see that there a lot of people involved in moving packages quickly. But we also see a huge amount of automation and you know that Amazon is trying to figure out ways to automate the remaining functions in these warehouses. It’s not a big stretch to envision robots taking the places of the ‘pickers’ in that video.

Some of the innovations on the horizon have the potential to eliminate other large piles of people. Probably the most obvious technology with that potential is driverless cars. One can envision jobs like taxi drivers disappearing first, eventually followed by truck drivers. But there are other jobs that go along with this like many of the autobody shops that are in business to repair car accidents due to human poor driving. We have already seen Starbucks trialing an automated system that replaces baristas and I saw one of these automated systems in an airport last month. There is a huge boom right now in developing manufacturing robots and this are going to replace much of the manual labor in the manufacturing process. But this also will allow factories to return to America and bring at least some jobs back here.

But this study predicts a much wider range of jobs that are at risk. The real threat to jobs is going to be through the improvement of cognitive software. As an example, IBM’s Watson has been shown to be more accurate than nurses and doctors in diagnosing illnesses. We are now at the point where we can bring supercomputers into the normal workplace. I read four different articles this week about companies who are looking to peddle supercomputing as product. That kind of computing power could start to replace all sorts of white collar and middle management jobs.

The study predicts a huge range of jobs that computers can replace. They include such jobs as patent lawyers, paralegals, software engineers and financial advisors. In fact, the paper predicts that much of the functions in management, financial services, computer technology, education, legal and media can be replaced by cognitive software.

Economists have always predicted that there would always be new jobs created by modernization to replace the jobs that are lost. Certainly that is true to some extent because all of those jobs in the Amazon warehouse were not there before. But those jobs replace store clerks in the many stores that have lost sales to Amazon. The real worry, for me, is that the sheer number of jobs lost to automation will happen in such a short period of time that it will result in permanent unemployment for a large percentage of the population.

One job that the paper predicts will be replaced is technical writer. As a technical blogger I say “Watson, the game is afoot! IBM, bring it on.”