I was in the middle of my comprehensive (end semester) examinations when the Google keynote was delivered. In all honesty I was dumbfounded after Sundar Pichai did that reservation demo on stage. I was dumbstruck again (x10 times the previous one) when Pichai went full devil mode and asked the google assistant to make sense of a thick, coarse Chinese-American accent. This feat is in no way less tremendous for all working AI softwares out there if not more than the moon landing human beings witnessed. It only broadens the scope to limitless possibilities in business, medicine , education and every other service based sector that you can think of.
Some key trend lines in the development of AI that we need to know are :
- AI market will grow to 153 billion USD by 2020 , according to the Bank Of America.
- AI start-ups venture capitalists have increased six-fold since 2000.
- Number of startups developing AI systems has increased 14 times since 2000
- AI will churn out 20% of content for business by 2018 end.
- AI is projected to create 13.6 million jobs over the next decade , says Forrester.
- Tech giants spent 30 billion USD on AI in 2016.
Now the main rhetoric against the use of AI in any process is that it is a JOB MURDERER. But take into account the thousands of AI start-ups and the number of people they need to keep the tech running is immensely huge. Thus the number of enrolments in colleges to study AI and ML has constantly been on the rise since the 90s. To put things into perspective here is a chart showing how the numbers have varied in Stanford University.(source : http://www.aiindex.org)
Introductory AI class enrolment has increased 11x since 1996. The share of jobs present in the market has also followed a similar trend. The following plot is plotted for data from monster.com . (source : http://www.aiindex.org )
With more and more people working on AI and more and more businesses integrating it into their systems , one cannot deny the fact that AI is here to stay and people will need and must embrace it to remain ahead of the curve.
Talking about learning curve, learning capabilities of AI have been a mixed bag and one cannot say for sure what to expect after the tech is exposed to data from the real world. While index has placed performance of the AI on specific tasks as excellent (beating Gary Kasparov in 1997 , AlphaGo Master took on Ke Jie and ruthlessly dominated him over three games of GO) , some experiences haven’t been really heart-warming to say the least. (Tay learned to be a racist asshole in less than a week on twitter). Some of the recent advancements in the performance index of AI has been shown in the charts below :
The findings are from competition data from the leaderboards for each LSVRC competition hosted on the ImageNet website.
Consumers have utilized more artificial intelligence than they are ready to believe in. Out of a sample set taken (6000 people) only 33% were ready to accept that they have in fact used an AI based service when 77% of that group actually used an AI-powered service or device. While only 38% believed that AI is going to improve customer service only 48% thought that AI would make the world a better place. We can observe the trend of the topics that are being published about AI. The following figure shows the keywords associated with AI in the articles published in The New York Times.
It can be easily inferred that the paradigm has shifted from all good , sweet sounding stuff to a kind of reservation and bias against AI. This interpretation however only represents one side of the coin. Even though the number of articles expressing concerns and being critical of AI have increased , so has the number of positive news from around the world as well. That should not come as a surprise as given the amount of funding going into the tech it is not very tough to influence people to write good about it and keep the market happy.
We can only trust the companies to take us to a place that prevents AI from being the last invention mankind witnesses. No government , no company can really save us from the havoc that AI might cause. However a common fear of losing out to the machines in the long run should prevent anyone from creating an AI powered system so strong that it becomes impossible to breach them. It could be like running against the fastest car we have built. We know how to operate it. We know how to stop it , but heck no we do not know how to compete with it. Simply putting my POV on this issue : AI is tremendously useful. It can prevent natural calamities , save hundreds of thousands of human and animal lives , make processes reach efficiencies that we have never heard of before but if we are to compete against it …… I will let Stephen Hawking do the talking here :(ironical since it was AI that actually let him talk)
It [AI] would take off on its own and redesign itself at an ever increasing rate. Humans, who are limited by slow biological evolution, couldn’t compete , and would be superseded.
PS : Featured Image from : www.extremetech.com