AI must be fair, accountable, open to explain its decision making: IBM India CTO3 min read . Updated: 30 Sep 2020, 01:21 PM IST
- Trust in AI comes only if it is built from ground up and not retrofitted. For companies to harness AI, they have to trust it. For that AI must be fair, accountable and open so people can explain how it went about making the decision that it made
NEW DELHI: Artificial Intelligence (AI) touches every connected individual and business in some way or the other. Post covid-19, the application of AI is expected to grow as businesses look to cut costs, increase efficiency and reduce turnaround time.
In a telephonic interaction with Mint, Subram Natarajan, chief technology officer, IBM India, sheds light on the growing interest in AI, how it is being used during covid-19 and why open and explainable AI is important for its future.
Q. Covid-19 has accelerated adoption of various technologies. Has AI based deployments also grown in India?
AI infusion has increased dramatically post covid-19. In every segment, including financial services, retail, industrial or automotive, most business engagements are happening online. Earlier, online engagements were limited but the pandemic has accelerated this. In the digital space, user experience matters a lot. AI plays a tremendous role when it comes to personalising a digital journey and tuning services specific to needs of a buyer. We have seen adoption grow in every sector. Even government has started adopting them for citizen services.
Q. How has AI been used in the fight against covid-19?
We helped ICMR create a virtual agent or chatbot powered by Watson engine. The chatbot is essentially integrated into ICMR portal in order to respond to specific queries of frontline staff. This includes data entry operators from various testing and diagnostic facilities across the country. It helped ICMR understand and respond to common queries in English and Hindi round-the-clock and as per ICMR guidelines. As a result, ICMR was able to automate data entry, allowing humans to be moved away from repetitive tasks to higher value areas. This was later emulated by various state governments.
Q. How are businesses in India using AI?
All our key customers are consuming AI in different modes, whether it is on prem or cloud. There are customers who are consuming simple solutions like email classification or ticket triaging that we have customised for different industries. Then there are large enterprises which are beginning to feel the need to establish a platform that allows multiple divisions from within their enterprise to derive benefits from AI and data. For instance, IFFCO Tokio General Insurance used AI to reduce turnaround time in claims settlements. We integrated AI with their claim damage assessment tool. It used cognitive image analytics to assess the type and extent of damage in lesser time. Similarly, for Parle we deployed an intelligent supply chain. The AI solution was able to predict the demand from different regions, reduce the time to market and right size the inventory for them. It reduced wastage dramatically.
Q. Is getting skilled people in AI still a challenge. What is being done to address this?
AI skilling has been a big impediment for different companies. To ensure this doesn’t delay adoption, we have a few solutions such as Auto AI which looks at data and suggests the AI model that you would need. It helps with quicker adoption of AI. Then there is advanced neural network, which allows practitioners to build neural network skeleton before they even begin. In terms of making sure there are enough skills available for future, we are working with various institutions and government of India. Recently, Central Board of Secondary Education (CBSE) integrated AI in its curriculum and IBM is providing the necessary content for it.
Q. There are a lot of trust issues with AI. How can it be addressed?
Trust and transparency in AI is a big focus area for us. Trust in AI comes only if it is built from ground up and not retrofitted. For companies to harness AI, they have to trust it. For that AI must be fair, accountable and open so people can explain how it went about making the decision that it made. It is extremely important so there is no bias in the models. This will allow ethical and fair use of AI.