For anyone who thinks the fruits of artificial intelligence (AI) are round the corner, wake up. It isn’t just caffeine-loaded programmers who can use AI. It isn’t just national security agencies that can use AI. Anyone can use it, and many are already doing it. There are specific areas within AI that are getting commoditised. You can go out and buy AI; maybe not every component, but enough to get you going.
So, what can you put into your AI shopping cart? At the moment, things like classifying content, identifying objects, facial recognition, and semantic search are just some of the areas that have become almost generic software services. Leveraging these,AI is being used to manage everyday things and improve society. We should celebrate that.
Just as everyday examples, traffic wardens are using AI to ease congestion on roads; the government is using it to intelligently deliver essential medical services , construction agencies are using it to figure out how much of the highway they have finished laying. A farmer could use a drone to take images of the field and get an accurate yield forecast. A tea-seller’s son in Siliguri, a town in Eastern India, has used it and hit the headlines. Ranjit Ghosh, the kid, who could not opt for science after leaving high school, built a drone that could identify dengue-causing mosquito larvae. Dengue is a life-threatening, pandemic-prone viral flu-like disease that flourishes in tropical climates. Ranjit's system uses AI to identify the killer mosquitoes.
Ranjit’s invention provides us an insight into what is happening around AI.
There are many algorithms, data models, data samples, and machine learning algorithms available that can be used by almost anyone. Even those who are not technology experts, can create disruptive services in a short period with minimal effort. Combining this with the fact that almost every industry and process has the scope to consume some cognition to improve itself, provides a scenario where AI concepts can be applied to improve and transform a very broad spectrum of processes. A field service engineer, for example, can use a mobile phone camera to photograph a faulty asset and let augmented reality (AR) technology quickly identify the part number of assets being handled, thus helping place accurate orders for replacements or get expert instructions on repairing it. A mining company can do stockpile estimation by analysing images captured by a drone. A factory supervisor can easily identify people who are not wearing safety gear in the factory. Implementing these use cases is simple with easy-to-use programming concepts and publicly available Application programming pnterfaces (API), training them with initial data and deploying them — something any programmer should be able to do. Creating simple chatbots using language APIs, creating speech to text interfaces using gadgets such as Amazon Echo are things even school kids are doing. The complexities involved in these areas which existed five years ago have been, to a large extent, abstracted and simplified. Everyone can let their imagination soar and make them yield exciting new things.
You just need the right AI components and assemble them, just like you would assemble a fire engine from Lego bricks. In fact, Lego’s own NXT-G development environment that comes packaged with the LegoMindstorm range, has been around since 2006, allowing people with no knowledge of programming to build devices powered by AI.
It has been, literally, kid stuff for over a decade! It is easy to dismiss this ability as “hobby” or “entertainment” or “education”. But the simple next step is to scale this sort of thing by using networks that can feed vast amounts of data to train AI. Even data is freely available. Cloud service providers, for example, have vast amounts of anonymised data available on demand for practically anything to get you started . You can build conversational or text based interfaces without investing a dime (admittedly, if you are serious, you do need to open your wallet a bit).
I reckon the time is right for AI to explode across enterprises in the next couple of years. AI is becoming simpler to handle and it is easier to see where to apply it within an enterprise for quick and maximum impact.
Clearly, the trick is not in mastery of programming or Big Data. The real challenge is to find uses cases that push the envelope. That’s how you will create the next Netflix.
To read the full story, Subscribe Now at just Rs 249 a month