Despite what companies like Microsoft may have told you, when it comes to truly understanding human language, computers are still just dumb machines. Sure, a computer can recognize the word "cat" and repeat it in a text message, but it has little concept of what a "cat" is. Whereas a human knows that a cat is an animal, a cat is a mammal, a cat is domesticated and humans keep them as pets, and that cat videos have taken over the internet.
This discrepancy between knowing a "word" and understanding a "word" is the missing link that keeps artificial intelligence and machine learning from reaching its full potential as a tool that can help users work, play, and live better. Notwithstanding self-aware machines taking over the world in an earth-shattering cataclysm, of course.
In November 2016, researchers at Microsoft Research Asia announced technology designed to help computers conceptualize in a humanlike fashion. They call it the Microsoft Concept Graph, and it may provide the means for computers to finally understand human language.
Microsoft Concept Graph
The Microsoft Concept Graph is a massive graph of more than 5.4 million concepts derived from machine-learning algorithms and billions of web pages and anonymized search queries. The idea is a major component of ongoing efforts to computationally simulate human thinking. If a computer can understand the concept behind a word, it can then understand the context surrounding the use of the word. So a computer can stop taking everything we say literally and instead start understanding and anticipating what we actually mean.
One of the practical concepts the researchers mentioned was keyword advertising. It would be helpful for all concerned if the keyword advertising you see when you are web browsing were more pertinent and a lot less ham-fisted. The subtlety of not only knowing that I am searching for the price of a house for sale down the block, but also knowing that I don't want to buy it and am just curious about the potential price of my house if it were for sale, is important.
With this kind of research, artificial intelligence may be able to make those subtle distinctions and provide a better experience for us all.
Artificial intelligence and machine learning applications are the prototype technologies that companies like Microsoft are trying so desperately to make an everyday reality. As a business strategy, it makes sense. Many of us are carrying computers around with us all day, after all. The more that smartphone can conceptualize and contextualize about you and the environment you exist in, the more it can do to help you navigate through that environment.
Research like the Microsoft Concept Graph uses our previous behavior patterns to give our computing devices "experience" they would not normally possess. Armed with that knowledge, someday your smartphone may end up knowing more about you than you know about yourself. Perhaps one day, all any of us will have to do in the morning is say, "Cortana, what should I do today?"
- Why AI and machine learning are so hard, Facebook and Google weigh in
- Machine learning: The smart person's guide
- Microsoft Cortana: The smart person's guide
- Bots give Microsoft Teams an edge on the competition—and on the future
Does the concept of a truly intelligent smartphone in your pocket make you happy or nervous? Share your thoughts with fellow TechRepublic members in the discussion thread below.
Mark W. Kaelin has been writing and editing stories about the IT industry, gadgets, finance, accounting, and tech-life for more than 25 years. Most recently, he has been a regular contributor to BreakingModern.com, aNewDomain.net, and TechRepublic.