Alibaba, which has built sophisticated machine learning algorithms to figure out the fastest and most cost effective way to ship a product from China to any part of the world, says it’s now willing to share this intelligence with other retailers. In India, Alibaba
Cloud, which wants to compete with Amazon
Web Services (AWS) and Microsoft, says it is doing just this and has begun working with its partners such as Paytm.
One of the reasons large companies have the edge in building these newer technologies, apart from their monetary might, is their access to huge volumes of data. In order to enable AI and ML systems to be accurate, there is a massive requirement for labeled sample data, which is readily available for firms such as Google
“The AI and machine learning algorithms and services are actually already commoditised, but the issue today is finding engineers who can use these services effectively,” says Lalitesh Katragadda, founder of Indihood and former Head of Google’s India Engineering Centre.
“What will happen over the next five years is that the algorithms will evolve to the point where the complexity of configuring them for learning a specific task will automate itself. When this happens, all you need is a services engineer, which we know how to find in plenty.”
Katragadda adds that the biggest challenge outside of these few large technology
behemoths is the availability of talent that can build such self-learning algorithms. The larger players, on the other hand, don't have ant computer science complexities to overcome as they go about building AI products and solutions that can be easily adopted by startups, enterprises and even traditional companies. All they need is a huge amount of engineering, which in any case is their forte.
While the day of picking AI programs off the shelf aren’t too far, there will still be scope for startups including those in India to do more innovative work in this space, says Umakant Soni, founding partner at Pi Ventures, a VC firm that exclusively looks to back startups in the AI space. For Soni, the talent shortage exists for non-tech companies that are looking to build AI tools by themselves, whereas entrepreneurs setting up AI startups are doing so to address the gaps they see.
“People might use AI on cloud services for doing prototypes or some kind of proof-of-concept works, but if you want to build robust solutions, you need to still do it on your own. AI is still not there yet and I think it will take 2-3 years to get vertical specific AI solutions from cloud providers. However, there will still be a need for specific AI solutions which other companies will be able to work on,” said Soni.
While Indihood’s Katragadda does see the large technology
giants dominating the space for AI and ML services, he says that things could change in the long run. As the processors in smartphones and laptops get more powerful going by Moore’s Law, much of the data processing that today happens on the cloud will begin happening on the device itself. This, he says, will be the vision in the next 10-15 years, but could stand to break any sort of monopoly that gets formed.