Watson: Artificial intelligence beyond Jeopardy
At a glance, artificial intelligence (AI) at its very essence seems to be just storage technology with big data and analytics solutions wrapped around it. In particular, IBM’s personification of its AI technology, Watson, has a layer of natural language added to it.
However, IBM Malaysia’s Chief Technology Officer, Freddy Lee and IBM technologist, Lee Yu Kit, share about an important element that differentiates AI from computers, known as probabilistic system.
Lee had described how computing had evolved over the years, and is on the verge of entering a new era of cognitive computing. “Computers are getting smarter, and you will see a lot of companies moving towards where computers are not just programmable, but able to learn.
The next logical step
The fact that we are creating more data in a year, than in the past 50 years, has necessitated this new kind of cognitive computing. IBM believes that by 2020, there would be 44 zettabytes of data in the world, that is equal to 1.6 billion years of HD video.
Also now with cloud computing, almost anyone can access information at any time, from anywhere via commodity devices that almost anyone can own.
But computers as we know them today, are not equipped to ‘compute’ these data. Today’s programmable computer cannot recognise a lot of data out there, Lee had emphasised.
That had led to something called big data analytics (BDA), which today is slowly and surely making its way into any business that generates and/or collects a lot of data. The use of BDA is to help make sense of all that data, and help businesses make better and more optimal decisions.
With probabilistic systems added into the mix, the idea is that computers would be able to learn.
IBM’s personification of artificial intelligence and what probabilistic systems are capable of, came into the limelight when it was pitted against human players in a few games of ‘Jeopardy’.
Since then, Watson has added capabilities to learn expertise from different domains and 35 Watson services have emerged for different industries like healthcare, retail and more. Offered as APIs that augment existing apps to make smarter recommendations, Lee said there would be nearly 50 Watson APIs by the end of this year.
Existing APIs now, enable applications with features like tone analyser, concept insights, sentiment analysis, personality insights, text to speech and more.
In Malaysia, Watson is offered as a cloud service, hosted on IBM’s on platform-as-a-service, Bluemix.