As smart as today’s conversational interfaces appear to be, when Alexa can’t distinguish what the word “it” refers to in a sentence, how valuable is she really?
Chatbots should be more than a party trick; and at this stage, many of their use cases leave much to be desired. But repairing their conversational shortcomings could create value where we never thought possible, and potentially be the engine that drives the IoT landscape forward.
The Key to Conversation: Context
Machines don’t understand context the way humans do, and solving this problem is the next step towards creating an interface that’s truly conversational. Luckily, progress in machine learning and natural language processing is giving new meaning to “conversational” devices. My colleague Katherine Bailey, Principal Data Scientist at Acquia, wrote an article recently about conversational AI and the road ahead. She talked about a machine learning technique called Word Embeddings, where vectors are used to represent words in 300+ dimensions, which introduces context into artificially intelligent systems. This technique is maturing, and conversational interfaces are getting smarter the more you speak to them. They can learn and categorize human language through experience, making them that much more useful for consumers.
To read this article in full or to leave a comment, please click here
June 09, 2017 at 12:04AM
from Chris Stone
No comments:
Post a Comment