The development of Big Data, which seems to have descended from the dark side, has impressive benefit.
Big Brother, the all-seeing invasive government of Orwell’s 1984, has a young grandson who is not nearly as sinister as Granddad. In fact, having gained a good deal of respect from his daddy, Artificial Intelligence (AI), the machine that thinks like a man, Big Data has grown up to become the delightful technology that empowers Google, Watson and the NSA, along with many other business analytical software applications.
Not everyone has forgotten the family connection, however. The lingering memories of Big Brother are still a trouble to some. But AI seems such an exciting and helpful technology, that most are willing to disregard the possibility that Robots might ever think better than humans (technological singularity). The latest serious prediction in this line is that human-level machine intelligence will come in the year 2029. Very close!
And while the AI branch of the family can easily win at Jeopardy (February 2011) and provide successful commercial medical ventures (September 2013), Big Data is growing in its ability to help AI out while carving a niche of its own in the business world.
To gain a better understanding of the role Big Data plays in AI, consider this commercial venture from IBM’s Watson. Medical literature grows so quickly that it would be impossible for even the best physicians to keep up with the information involving their own field of expertise let alone the entire medical database. Some estimate that only 20% of what the average doctor prescribes or performs in his practice is based on real medical evidence backed by new scientific data.
For example, the American Cancer Society projects that in one year in the United States 1.6 million new cancer cases will be diagnosed. In how many of those cases will the diagnosis accurately reflect the latest understanding of the particular kind of cancer diagnosed? There are studies that point to the complexities associated with healthcare as having caused 20% of healthcare patients to receive a wrong or incomplete diagnosis.
Medical information is doubling every five years. The doctor who doesn’t keep up is more prone to be a part of the increasing percent of those who make a wrong or incomplete diagnosis. These statistics, coupled with the data explosion of medical information, represent an excellent opportunity to see the business benefits of Big Data. The healthcare industry will be able to make use of next-generation AI computing systems. The analytical use of such extensive data will go beyond accuracy in medical practice to take a step toward how the practitioners are trained.
IBM entry into the world of AI is called Watson. IBM has taken their expertise in AI to partner with two medical institutions, WellPoint and Memorial Sloan Kettering. They have used these relationships to gather data that will, in effect, train Watson in the areas of oncology and utilization management. In fact, clinicians along with technology experts have spent thousands of hours in organizing and analyzing Watson’s data. This effort has enabled Watson to more efficiently process, analyze, and interpret the meaning of complex clinical information.
At this particular point – working through the data analytically – we have only approximated the nature of Big Data. But Watson has the particular AI facility that allows users to interface with the machine using natural language processing. The end goal is to help improve healthcare quality and efficiency. Presently, 90% of nurses who have access to the data follow its guidelines.
Imagine being admitted with cancer and having your oncologist query Watson to come up with the best diagnosis to follow. In moments, billions of bytes of data are compared against the particulars known in your case to provide clear next steps. The doctor, who may not have had opportunity to read the most recent 10,000 pages of literature identifying various symptoms and solutions, is able to glean the best of the best in moments. You benefit! As does the hospital. And when the doctor inputs your treatment and health progress, so will Watson.
Another instance of Big Data fueling AI is Google Translate. Anyone can use this software online for free to translate text from one of 70 languages to another of any of those 70 languages. If you aren’t sure what language it is, Google will guess (accurately) and translate.
The Big Data side of Google Translate is a database of billions of word combinations mapped between two or more languages. So what you type (or write) in English has a corresponding translation in Russian (or French). And on writing the Russian, you can get the English (or Spanish).
The AI side of this tool kicks in by figuring out sentence structure and meaning without being told what the verb or the noun is – no grammar is required. The software system is able to figure out what is what. AI in Google Translate works without a theory being given. The process is data driven and all the combinations are brought into play based on an algorithm – rules that are part of the software system and too complex for most humans to understand. The correlative algorithms enable Google Translate to get better over time. In essence the system is acting as humans do in the process of learning.
Another Big Data facility at Google that is seeking to approach the level of AI is Google Vision. Google Vision has 16,000 microprocessors that specialize in computer evaluation of images (remember that every YouTube video is a collection of images, like the old movie frame). This aggregation of microprocessors is only 10% of our brain’s visual cortex. A lot of training still needs to happen. Billions of billions of still images from YouTube videos and Google Images are a part of the process.
In a recent experiment, Google Vision studied images of cats for three full days. During those 72 hours the AI of Google Vision learned how to recognize the image of a cat. Not with 100% accuracy, as there are still problems with size, type, color and fur of the cat. Still Google Vision sees twice as well as any other computer. Compared to a little child we might note that a child has greater accuracy, but that AI has no fear!
On Watson’s website you can read of the other projected accomplishments for the AI machine.
- Help planners recommend better investments – Reuters publishes 9,000 pages of financial news every day. Watson can collect not only news, but analyst documents, e-mails, transaction data, and public opinion to help investment advisors confidently recommend investment and portfolio changes
- Help operators answer questions more effectively – Call centers leave 135 billion matters unresolved each year. Almost two-thirds of them could have been resolved with better access to information. Enter IBM Watson Engagement Advisor –- technology that can listen to caller queries and suggest follow up questions, and that can help operators find answers faster
- Help developers get disruptive – Say hello to the Watson Ecosystem: the IBM Watson Developers Cloud is a cloud-hosted marketplace that offers technology, tools and resources to help developers build cognitive apps and bring them to market. The Content Store will offer general data and industry-specific content for use by developers to fuel their cognitive apps, and the Talent Hub will offer staffing with skills like natural language processing and machine learning
- Help researchers find info faster – The Economist estimates companies spent $603 billion in research and development in 2012. That’s a lot to look up. The IBM Watson Discovery Advisor is a research assistant that helps researchers collect information and synthesize insights to stay updated on recent findings, and share information with colleagues
Big Data has a Little Sister, by the way. She’s called, Internet of Things. Check out our article posted on this fascinating subject here.
