Original article can be found here (source): Artificial Intelligence on Medium
Alexa in India
According to Mr. Prem Natarajan as said by him in the keynote of voxcon India the major challenge for the company is to make Alexa more locally relevant by locally relevant I mean in India majority of consumers are multilingual so while talking to Alexa if a customer uses the word “gum” then Alexa should be able to recognize that whether he is referring to the Hindi word that means “sad” or the “chewing gums”.
Mr. Prem also emphasized the point that adding Hindi to Alexa was itself a very challenging task because hindi as a language is used by people here in India in multiple ways and the tone and pronunciation changes very drastically as we go from one place to other. The most interesting thing that I found when it comes to consumer behavior here in India was that the families itself are very diverse as in a family of three members the husband is from Chennai, the wife is from Delhi and their child is born and brought up in Ahmedabad this family will be using same Alexa device and all three of them have different accents which itself is a challenge for the device which work on conversational AI. There are a million such use cases in India.
You can watch Alexa ads on youtube to see different use cases the company is targeting while watching those ads I would encourage you to think as a developer rather than a consumer and you will see all of them show completely different dynamics.
There are 5 pillars on which Alexa is functioning and becoming more and more relevant to the people –
- Context Awareness
Alexa is becoming more and more context-aware i.e its understanding the intents more properly example:-
You- “Alexa, How’s the weather today?”
Alexa- “Sunny 27-degree Celcius”(something like that)
You- “What about tomorrow?”
Note here in the above dialogue when you didn’t mention the context in the second question that you are asking about the weather but the Alexa is context-aware i.e she knows this is a discussion about the weather in Delhi so she gives the appropriate answer.
The vision is to make Alexa act like one of the family members so it is much more important for her to act and behave like one.
In the same event voxcon India here’s a dialogue between Mr. Dilip RS and Alexa.
Dilip- “Alexa kya haal hai?”
Alexa- “Bas aapne puvh lia dil garden-garden ho gya!”
This is the type of Indian touch company is aiming for, of course, you will never accept an Alexa device which pronounces Swami Vivekanand and Sachin Tendulkar the way President Donald Trump did on his recent visit to India.
Self-learning is the basic pillar of any AI-based device, in the case of Alexa, we can discuss some use cases as when you say “Alexa book me a cab” Alexa should be smart enough not to ask any further questions like “ola or uber?” she should book a cab by analyzing your personal preferences she already has.
Another case is when children ask Alexa to “play ABC song” instead of “Alphabet song” and it is corrected once or twice. Now Alexa is smart enough to play the Alphabet song whenever the next time there is a request for ABC song.
When it comes to a device that has to be operated worldwide the knowledge about local contexts is very important as in India if someone is asking score from Alexa then it is pretty sure that he is asking about cricket scores. The answer to some questions is different for different places in the world. For achieving the dream of an Artificial General Intelligence the systems have to be made as knowledgeable as they can.
Competence means the capabilities to perform tasks. Alexa is becoming more and more competent, this is due to the efforts of the developers worldwide who made almost 90000+ skills and out of these 30000+ skills are available in India.
Alexa is now also able to perform multiple intents like “Alexa play pop music and dim the lights.” in this statement two different commands are given at a time and Alexa is now competent enough to perform both of them together.