Who’s sick of hearing “I didn’t quite get that” or “I don’t know how to help with that” from their virtual assistant or smart home device? I know I am, so when I saw the recent news that Alibaba Group’s AI achieved a higher score than humans in a well-known test that has been developed by Microsoft, I had to sit back for a minute. I think that it’s amazing to see the pace at which AI is improving and what that means in the future for us.
The Microsoft Machine Reading Comprehension test is used by machine learning industry as it is difficult and really proves a machine learning’s systems are capable. The test contains a dataset of real Bing search engine user queries that are used in reading and comprehension, question answering, passage ranking, keyphrase extraction and conversational search studies.
The Alibaba model scored 0.54 in the question and answering task. In comparison, the best human score was 0.539 which has been a benchmark provided by Microsoft.
This might seem like a fancy experiment and the above doesn’t mean much to you but it truly is important. Any AI system’s ability to detect nuance in text and speech allows it to “understand” better and in the end provide you, the user with contextual answers that are actually correct and relevant. This step forward will see huge implications for the way we converse with our smart assistants and their ability to answer better.
It’s not only virtual assistants that will be improved; think about the upcoming Chat Bot revolution that is happening? They are only starting to serve their usefulness but with this, they will truly provide a utility and no longer be simply a storyboarded question and answer solution.
Alibaba currently uses its AI platform within its customer service bot. During its 11.11 festival – it’s largest shopping day of the year – they were able to serve 50 million active users. Their AI bot was able to handle 98% of all requests made that day, and at its peak, it was serving 83,000 customers a minute.
AI has been a huge buzzword for the last few years and examples like this really show that it is progressing and no longer in the realm of Sci-Fi. When a platform can serve 98% of queries to the largest online store in the world, it has become a standard operating procedure and allows businesses to scale at an unprecedented pace.