Machines getting more intelligent


With democratisation of AI, developers are able to make better use of niche technologies

Imagine you applied for an insurance cover and uploaded all the required documents. After a few days, you get a call, and you are told something on these lines: “Thank you for submitting the documents. However, the income proof you submitted is incorrect. Can you please send us your 3 months’ bank statement or 3 months’ salary slip?”

Nothing unusual in this, except that it was not a human being at the other end of the call, but Amazon Polly, a text-to-speech service. Such messages are generated by the software that is able to not only scan documents to make sense of words and numbers with a high degree of accuracy but also render instructions like a human being.

With Artificial Intelligence and allied technologies vesting machines with more capabilities, automated messages are now more personal and specific rather than generic. Earlier, the previous automated call would have been on the lines of: “Thank you for submitting your documents. However, there is something wrong. Please contact the help centre.”

PolicyBazaar, for example, after incorporation of Amazon Polly into their IVR calling service, PB Connect, is reporting much better engagement with their customers, especially in cases where there is a specific issue that needs to be resolved. “As many as 80% of the calls made by PB Connect were answered, 53% responded positively, and 41% of sales were closed without the interference of agents,” says Ashish Gupta, Chief Technology Officer, and CEO

Cloud computing services providers are increasingly tapping into machine learning capabilities to help businesses find more precise solutions. Developers are able to provide more intelligence to their applications, with a better analysis of text, images, videos, and more accurate recommendations and predictions.

“We are democratising machine learning capabilities by putting them in the hands of developers and data scientists,” says Olivier Klein, Head of Emerging Technologies, Asia-Pacific, AWS. “Now, any developer can make use of these services without having to understand any machine learning and without having to train the models. At the recent AWS re:Invent, as many as 13 new machine learning capabilities and services across all layers of the ML stack were launched,” he said.

AWS has launched more than 200 significant machine learning capabilities in the past 12 months. Among the many customers using them are Adobe, BMW, Cathay Pacific, Dow Jones, Formula 1, Johnson & Johnson, Shell, Tinder, United Nations and the World Bank.

Using Amazon SageMaker, developers are able to get their models to production quickly without much effort and at a lower cost, since they have a fully managed service that covers aspects like preparing data, choosing the right algorithm, training, tuning and optimising it for deployment, making accurate predictions etc. GE Healthcare, which manufactures and distributes diagnostic imaging equipment, runs its GE Health Cloud on AWS and makes use of SageMaker.

“We have been investing in AI and ML for the past 20 years. We have thousands of engineers focussing on these technologies, and we have the largest and deepest set of portfolios of AI and ML,” says Navdeep Manaktala, Head of Business Development, AISPL.

An AI-based solution that identifies text, images, people and activities besides inappropriate content is Amazon Rekognition. Matchmaking website, for example, after adopting this solution, has done away with the practice of human intervention to review the images posted by the users. Besides, Rekognition has been able to reduce the time taken for the photos to appear in the profile.

Another application of Amazon Rekognition is in tracking missing persons. The ReUnite app, which was launched in June last year, uses the Face Search feature to match photos of missing children with a database. The solution comes up with even the extent of similarity between the images. This technology also finds application in other areas like authorising employees to access secure areas.

AI is now used extensively in data management. Not only are there multiple types of data, but some are used very often while others are not. Also, some data are sensitive but others are not. Amazon Macie uses ML to protect data more securely. It detects risks and suspicious behaviour, provides alerts and recommendations on how to solve issues.

Why you should pay for quality journalism - Click to know more

Related Topics Gadgets
Recommended for you
This article is closed for comments.
Please Email the Editor

Printable version | Jan 26, 2020 9:07:15 PM |

Next Story