Microsoft and Alibaba officially have AI that’s smarter than humans
Microsoft and Alibaba are the first two firms to officially beat humans using the Stanford Question Answering Dataset, known among researchers as SQuAD, reports Telecoms.com (FinTech Futures’ sister publication).
SQuAD is essentially a reading comprehension test for artificial intelligence (AI) algorithms which uses questions based on various Wikipedia entries. University researchers and various businesses around the world can use this test and receive a score to see where their algorithm stands up against the rest of the world.
Over the last couple of days, both Microsoft and Alibaba put their AI up against the team and did very well. Microsoft got a score of 82.650, while Alibaba managed 82.440. Both of these scores beat the human performance of 82.304 for the same test.
“Microsoft has made a significant investment in machine reading comprehension as part of its effort to create more technology that people can interact with in simple, intuitive ways,” says Allison Linn on the Microsoft blog.
“With machine reading comprehension, researchers say computers also would be able to quickly parse through information found in books and documents and provide people with the information they need most in an easily understandable way.”
This is certainly a positive step forward for the AI world, as while it might seem like a simple test it does demonstrate the potential in customer services. For customer interaction platforms which are text based (as opposed to on the phone), Microsoft and Alibaba have pretty much nailed it. Such scores indicate the human operative can be replaced, in theory, without any negative impact on performance.
Having a look through the SQuAD leaderboard is certainly an interesting read, and while we would not want to take the shine off some very good Microsoft news, there are a few missing names. Two names which we suspect would perform quite well at this task would be Google’s Deepmind and Amazon’s Alexa research team. Microsoft and Alibaba are certainly making progress, but it would be interesting to see where Google and Amazon would figure on this list.
The achievement from both firms just demonstrates how close the industry is to actually nailing AI. It has been gradually gaining greater traction in our lives over the last couple of months, but most would assume a radical disruption from AI would still be years away. Perhaps this leadership table just indicates how close the day is when machines will take jobs away from people.
For those who are beginning to look over their shoulder, there are still a couple of hurdles saving our race. Firstly, this is a text system, there will still be millions upon millions of people who will want to actually talk to someone. AI might be able to handle the text interactions, but there will at least be a job for phone operatives for a while.
Another area is on natural language processing. The majority of these Wikipedia pages will be written in standard, grammatically correct language; this is not necessarily how a customer will interact with a company.