India to play key role in EY's push for proprietary AI language models

EY will leverage its global delivery unit in India, home to most of its tech staff, to lead development of proprietary language models for tax, audit and accounting services

India can become a critical cog in the wheel for EY to build its proprietary language models that will help in areas such as tax, audit, and accounting, two senior executives from the firm’s global delivery business (GDS) said.

 

Just like other areas, tax and accounting are not immune to the impact of artificial intelligence (AI). Enterprise language models, trained on client or synthetic data, can help accounting firms perform tasks easily and thus improve efficiency and productivity.

 

“Our intent is to build a large language model and we started in that process,” Raj Sharma, global managing partner, EY, said in an exclusive interaction with Business Standard. 

On the progress of the model and if the firm has met certain milestones, he said EY was in the first quarter of a four-quarter game, using an American football analogy.

 

EY GDS is the backbone of EY, providing services from technology and operations across the breadth of the organisation. It has about 60,000 employees out of the 80,000 globally in India. GDS is present in eight other countries including Argentina, Mexico, Poland and The Philippines.

 

To bolster its technology backbone, the company has also set up an AI factory or a centre of excellence in India, along with the one in Palo Alto, California.

 

“We wanted to make sure that we do not use India just as a labour arbitrage on these new services. There is a shortage of talent all over the world when it comes to AI and others and our strategy was to start building it from the ground up in locations that make sense,” added Sharma.

 

The other areas are blockchain, sustainability, and engineering. For the last vertical, a large part of product innovation and development is done out of India for its business-to-business Cloud platform.

 

“The domain knowledge that is required to create the large language model will come from all our member firms. We will bring our people. The engineering aspects, analysing the data, cataloguing the data, doing the engineering work and tuning the models to be able to create — in a lot of that, India will take the lead,” said Sharma.

 

For EY GDS, India is a natural fit to lead building these models because a large part of its technology team sits in the country.

 

Around 23,000 of the 60,000 here are with a technology background.

 

“One of the advantages of a global capability centre (GCC) is you have local knowledge in functional and technical. I have 140 member firms. The only place where it comes together at scale is here. I have domain knowledge, people who do the work, people who are filing all those taxes. I have a platform that is built out here and I have my technology team that is out there. Those are three great ingredients of fulfilling this four-quarter game that we are trying to play,” added Sharma.

 

“What we have in our all-in strategy is deciding to do the development here because once we do it here, we can take it to all those member firms rather than each individual firm,” Ajay Anand, global vice-chairman of EY GDS, said.