Is AI Friend or Foe? Experts Discuss Trust, Transparency In The Age Of Generative AI
From autonomous vehicles to personalised recommendation systems, the boundless potential of Artificial Intelligence (AI) has ignited a global fervour of excitement and anticipation. Yet, within this fervent optimism lies a tapestry woven with complexities and concerns.
As industry experts gathered at the BW Businessworld Gen AI Summit 2024, a panel chaired by Wg Cdr S Sudhakaran (Retd), Managing Director & Chief Executive Officer of QuGates Technologies, the discourse veered towards the ethical dimensions, privacy concerns and the imperative of trust in the age of Gen AI.
"It has become important to know where the gen AI tech is heading, is that meant to have the kind of attention it’s getting today or is it an end of all or is it something more sinister to come out and how it’s going to affect ourselves," remarked Wg Cdr S Sudhakaran (Retd), setting the stage for a deeper exploration.
There is been a lot of hype about AI, and the expectations of what can happen with AI are huge and there is a big gap between these two, Sanket Atal, Managing Director, India Site, Operations and Technology & Product, Salesforce India explained.
"There is a person responsible for executing like that within an enterprise, who got to make sure that the latest and the greatest technologies are used to serve the enterprise but at the same time it’s done in the manner that conducive to business and its growth and there is a big gap between these two, what people expect will happen where people are today with respective what will satisfy them to be adapted the AI truly can be a reality everywhere," Atal emphasised.
Shruti Kharbanda, Founder of Invincible Robo League Robotics & AI Chair (Northern Region) at the Indian Chamber of Commerce, noted the dual nature of AI's capabilities.
"Generative AI is an amazing technology, we can generate new images, text, audio everything, it is a powerful tool. But with the powerful tool, there is another side to it," she said. Beneath the Gen AI, the algorithm, they all learn from the data and that is where people need to take a step back and start focusing, training data and real data that is being used.
She emphasised that with the maturity of AI, the lines between the virtual and real world are worrying. "The oversights, the hallucinations, that need to be recorded and rectified, the omissions that have to be regularly identified, the errors in data that need to be rectified, and the biases that need to be removed," she added.
Delving deeper into the discussion, Narasimha M, Vice President & Head Data Science SBU at MakeMyTrip, outlined key factors essential for fostering trust among customers.
"Some of the large language models are foundational, general-purpose models. There’s a lot of knowledge also encoded into these models, besides language and conversational abilities." he said.
He cautioned against relying solely on large language models for knowledge, highlighting the potential pitfalls of such an approach.
"If you rely solely the knowledge encoded within these languages, you’ll face surprises at one point or the other. Instead, modify your genAI systems architectures, such that you hook into your datasets or reliable sources for facts, while using large language models to converse, reason around, or articulate those facts. This gives you opportunity to be transparent increase trust among customers" he stated.
