Why India risks falling behind in the AI ​​race

Why India risks falling behind in the AI ​​race

India’s tech industry is not being bold in embracing artificial intelligence. He hopes to build solutions for enterprise customers based on someone else’s investment in core technologies, which is hardly a strategy for pioneering success.

The high-voltage debut of ChatGPT last year galvanized China. Baidu’s Ernie, who claims to have outperformed the Microsoft-backed OpenAI model by some measures, has drawn Ant Group and JD.com into the bot-building race. Tech czars like Wang Xiaochuan, founder of search engine Sogou, have also joined the search, luring talent into the industry. On cash flow, the US is still leading China six to one, but the number of venture deals in the Asian country’s AI industry is already outpacing consumer technology, according to data from Preqin.

India’s startup landscape, meanwhile, is stuck in a time warp, with constrained investors slashing their stakes in Byju’s, an online education company that is collapsing under the weight of its own reckless growth. Easy pandemic-era funding has dried up. As funders press founders for profitability, they discover that, in many cases, even the revenue is fake.

This was the perfect time for traditional Indian coding powerhouses – such as Tata Consultancy Services and its rival Infosys – to put their superior financial strength to use and assert leadership in generative AI. But they have their own governance challenges. TCS is distracted by a US job bribery scandal that it is desperately trying to play down. Infosys is busy running the scam of its association with an Australian lobbying firm at the center of a parliamentary investigation into Down Under.

Even without these challenges, outsourcing specialists aren’t exactly in a sweet spot. Demand for its services is weak, mainly because of the turmoil in the global banking system. IT spending decisions have declined. Tougher competition for a smaller pie could mean lower orders and a deterioration in prices, analysts at JPMorgan Chase & Co. earlier this month. Meanwhile, payrolls at Indian companies are bloated, thanks to the hiring spree during the pandemic as clients struggled to digitize their operations.

It’s no wonder that the industry’s approach to AI is defensive, geared toward assuring investors that the technology poses little threat to its time-tested model of labor cost arbitrage. When three lines of C programming replaced 30 lines of assembly language, it didn’t lead to mass layoffs, but to an explosion in code writing. Likewise, when outsourcing made enterprise software cheaper, IT budgets didn’t shrink. Volumes increased as prices dropped. Why this time should be any different, asks the TCS annual report for 2022-2023.

This is a rather phlegmatic reaction to a revolution whose possibilities are beginning to frighten its own creators.

ChatGPT can certainly write code snippets or perform a quality check on them, potentially reducing billing hours. But this is not the point that needs to be addressed. Being around machines smarter than any of us holds troubling prospects for the future of humanity, especially if algorithms come to be controlled by evil actors. Even setting aside these deep concerns about a potentially dystopian future, the more prosaic issues matter to enterprise software users as well. Companies from banking to retail to aviation must decide whether to engage with so-called broad language models. And they can’t be sure that taking something off the shelf is good for data privacy. What exactly are Indian companies doing to seize this opportunity?

Bengaluru-based Infosys has adopted a mix-and-match strategy, so its customers can choose from 150 pre-trained models on more than 10 platforms and then run them on any cloud or internal servers. The TCS annual report says that its research into large language models is geared toward “creating techniques for controlled code generation, answering questions, consistent image generation, solving optimization problems, and other core AI problems.”

However, if Alphabet is warning employees about how much information they can share with chatbots, including its own Bard, then how can TCS or Infosys assume that global multinationals will be comfortable pitching their tents on platforms available to virtually anyone?

Indian software services companies should also be building language models from scratch for themselves and their customers. Yes, it takes computational power and engineering talent to train neural network-based programs on large amounts of natural language input. But not going that route and looking to connect customers through application programming interfaces, or APIs, to existing products is unnecessarily timid, especially when no serious company can want to rely on a publicly available external foundational model for mission-critical tasks.

Google’s own research into extracting training data, or the potential for models to leak details of the data they are trained on, shows that the risk is very real.

Building proprietary and well-protected core technologies is not resource-intensive. For Nvidia co-founder Jensen Huang, whose chips are at the heart of the AI ​​excitement, even a modest budget of US$10m (about Rs.82 crore) for full-scale models is not unreasonably low. Countries that are not traditionally known as technology producers are also being noticed for their innovations. Abu Dhabi’s Technology Innovation Institute has made its Falcon 40B – trained on 40 billion parameters – royalty-free for commercial use.

The Chinese are clearly not buying the idea that Silicon Valley will control the keys to generative AI. While the excessive service orientation of Indian software companies has meant very little success in product development, now is the time for some ambition and a new strategy that goes beyond charging customers a fee to tweak OpenAI’s GPT-4, Google’s Bard or Meta Platforms’ LLaMA with specialized data.

On a recent visit to the country, OpenAI CEO Sam Altman was asked if someone in India with $10 million (about Rs. 82 crore) to invest should dare to build something original in AI. He said: “The way this works is we’re going to tell you it’s totally pointless competing with us on training foundation models (so) you shouldn’t try it, and it’s your job to like it, try it anyway.”

The message from Abu Dhabi is very clear: Bengaluru should try anyway.

© 2023 Bloomberg LP

Affiliate links may be generated automatically – see our ethics statement for details.

Leave a Comment

%d bloggers like this: