AI is probably not the end-all-be-all solution to all things in life, after all. Even when it comes to conversational experiences like Amazon Alexa, human-solved problems are sometimes way better to machine-solved ones.
According to a recent discussion held by Janet Slifka, Amazon Alexa’s AI director of research with Venturebeat, she reveals something along the same lines.
She went on to say that whenever someone calls the customer care and goes, “Hey, Alexa can’t understand me,” — it’s much more convenient to manually come up with a fix.
On the other hand, relying on machines or AI to do so is slower, or more laborious from the customer’s perspective. Also, the constantly changing colloquial language is also a factor.
She confesses that people can’t be waiting around for technologists to build a statistical model every time a new word gains prominence. That would be something not-very-feasible.
This commitment towards consistence is showcased by Alexa Answers, a simple approach to expand Alexa’s knowledge base.
What this platform does is that it allows customers from the US and Germany to submit answers to their Alexa-unanswered questions. Amazon claims that this helps Alexa answer questions faster than an automatic backend ever could.
Slifka also said, “When you start with the customer, you look at all of your tech options. You may not choose the latest and greatest that caught your eye.
You might allow that the newer things that you’re doing are messier and more challenging. For that reason, you might have more manual processes in there until you really hone things.”
1. 8 Proven Ways to Use Chatbots for Marketing (with Real Examples)
2. How to Use Texthero to Prepare a Text-based Dataset for Your NLP Project
Amazon is set to apply more autonomous AI-driven solutions to the problems in the future. Scientists at Alexa AI are trying to learn supervised and unsupervised learning.
Meaning, the AI is going to undergo training such that they can make predictions without cramming themselves up with annotated data. One such example is Alexa Conversations, which allows developers to create Alexa apps and also interconnect it with other apps without any hassle.
Another great example is Alexa’s smart home integration system. If people ask her, “Alexa, turn on the sofa lights,” but the lights are named as ‘living room lights’, then she’ll automatically suggest, saying, “Did you mean living room lights?”
The whole idea is to remove the need for a manual fix and let the AI do it all. Because let’s be honest, the statistical model is a stronger option for the customer any day.
Seeing the strides tech giants are making, and the new problems and solutions they are coming up with, its no news that soon we’re heading into a technologically refined world. It shall not be a surprise if, in the future, we have AI suggesting questions alternatives, or deciphering what we actually try to mean. And we owe it all to voice technology!
