InterConnect: Mythbusters – how Watson works

Share this post:

Have you ever wondered how Watson, IBM’s AI works? Attendees at IBM’s InterConnect 2017 were able to find out from Rob High Jr. IBM Fellow, VP, CTO Watson, IBM Academy of Technology. Rob kicked off by defining the characteristics of what IBM sees as AI before busting a myth or four.

  1. Cognitive systems understand human expressions – textual, verbal, visual
  2. By reasoning about the actual intention or problem being addressed
  3. They learn how to recognize patterns of meaning through examples and feedback
  4. And they interact with humans on their own terms, and in a way that inspires people.
  5. And they do it at scale!

Myth #1 Watson is only used by super large companies that have super-sized computers

True, Watson is used by super large companies that have super-sized computers. But that’s just a part of the many types of organisations and people who use Watson. Watson has a number of components. Platform as a Service along with developer tooling. Software as a Service along with Application tooling. Skills as a Service with maker tooling and Data as a Service with Content tooling. Watson can be delivered on premises – which brings to mind super large companies with super-sized computers. But Watson can also be delivered in the cloud, or a hybrid of the two approaches. Data can be public, private or even crowd sourced – opening the use of Watson to significant communities.

Myth #2 Only super-PhD types can build applications using Watson

Watson is built up of a number of API’s. They nestle under the four categories of language, speech, vision and empathy. Within language is a set of capabilities centered around understanding text and copy such as taxonomy, keywords and sentiment analysis. In speech are the capabilities that translate text to speech and vice-versa. There’s also keyword spotting and telephony speech to text. When it comes down to vision, Watson can spot celebrities with celebrity recognition, classify images and detect faces and attribution. Lastly there’s empathy where Watson has tone analysis, emotion analysis and can provide personality insights. This list isn’t exhaustive, Rob explained that all the API’s can be found by searching for Watson services.

Watson’s tone analyzer for example uses psycholinguistics, emotion analysis and language analysis to assess tone. When you’ve a raft of customer feedback content to sift through, it can swiftly surface the sentiment of your customers and point you towards areas such as joy, anger, disgust, fear and sadness in their feedback.  With Personality Insights, Watson can comb through your content and let you know how likely a customer is to click on an ad, follow on social media or buy eco-friendly (among other actions).

Myth #3 Watson can only be trained by mad scientists

Rob reflected back to the third element that defines AI systems, their ability to learn how to recognize patterns of meaning through examples and feedback. To learn, Watson needs to be taught, and that’s done through Watson Knowledge Studio. This enables subject matter experts and developers to teach Watson the linguistic nuances of industries and knowledge domains. The more input Watson receives, the better it gets at providing insights – and that input can come from anyone with domain knowledge, not just mad scientists.

Myth #4 Watson wants to take over the world

Watson wants to have a conversation. Gone are the days of ‘press 1 for this, press 2 for that’. Today’s customer assistants are more likely to be chat bots or virtual agents that use a natural language interface with their customers. Able to work on any messaging platform, these conversational characters are trained on user defined intents, entities and dialogs. They can be expanded to recognise and respond to the user’s emotion. Rob explained that there are four elements to a successful virtual agent or chat-bot:

  1. Engage the user
  2. Focus on the user’s broader concern
  3. Build on an idea
  4. Leave the user inspired and satisfied

If you’ve ever asked Siri or Alexa a question, you’ll know first hand how these conversations feel.

Myth #5 Who is the voice of Watson?

1,000’s of people tried out for the voice of Watson. Gone are the monotone voices of old. Now Expressive SSML and Voice Transformation SSML bring life and a human lilt to computed voices. Expressive voices can be directed to sound apologetic, uncertain and sympathetic. Empathic understanding can help guide a response in line with the customers emotions, that is in line with that feeling. Glotal tension, timbre, breathiness and pitch are among many elements that can be uniquely configured to provide just the right sounding response to any situation. Rob gave some examples of just some of the Watson enabled voices that you may have interacted with: tax advice with H&R Block, personalised weather with the Weather Channel even wedding planning with Meeka.

Watson is part of IBM’s IoT offerings. You can find out more in the ‘7 Deadly Sins of IoT Strategy and Design‘. Or stay apace of all the latest news by signing up to the IoT Sense monthly newsletter.

More Asset Management stories

Top Takeaways from IoT World 2019

Written by Sarah Dudley | May 20, 2019 | Asset Management, Manufacturing, Platform

The total number of Internet of Things (IoT) devices is expected to hit 10 billion by 2020. With this explosion of growth also comes an uptick in the conversations around how to use these devices and the data they generate to improve our businesses and our lives. At IoT World, North America’s largest IoT event more

And the winner is……Watson IoT at the IoT World Awards 2019!

Written by Sarah Dudley | May 17, 2019 | Asset Management, Platform

It’s been an exciting week as more than 12,000 people converged on the Santa Clara Convention Center for Internet of Things World 2019! Topics have spanned from connected cars and smart buildings to Industry 4.0 and developing an IoT Platform. In the full recap going out in the next few days, we will highlight the more

Do you know the price of poor infrastructure?

Written by Stephen Russo and Bruce D Baron | April 24, 2019 | Asset Management

Over the last 20 years, huge investments have been made in metropolitan areas to understand the immense volume of data collected about the citizens inhabiting these cities. Hundreds of millions of dollars have been spent combining data with emerging mobile, sensor and machine learning technologies. This is all done with the goal to improve infrastructure more