Unlock Editor's Digest without cost
Roula Khalaf, editor of the FT, picks her favorite stories on this weekly newsletter.
Indian tech grandee Nandan Nilekani predicts that firms all over the world will increasingly develop their very own smaller artificial intelligence models to streamline operations and increase productivity, raising hopes of serious gains for more powerful generative products in dampens the corporate.
The chairman of IT services giant Infosys told the Financial Times he’s “not so sure” whether firms wish to shoulder the high costs and potential “black box” of knowledge and copyright liabilities related to large language models behind popular applications resembling OpenAI are ChatGPT.
“If you take a look at the large firms, they're all asking themselves, 'How will we take control of our AI destiny?'” Nilekani said in an interview in Bengaluru, India's Silicon Valley. “Small language models trained on very specific data are literally very effective.” . . Everyone will construct models, but I don’t think they should construct these gigantic models.”
Nilekani said corporate customers would seek LLMs for specific applications and the technology would fuel competition between giants like Google and Apple that provide online products to consumers.
However, his prediction raises questions on the revenue model for startups which have invested in LLMs. Amid corporate hesitation, firms like Apple, Google, Meta and Microsoft have all recently released AI models with fewer “parameters” – the variety of variables used to coach their systems and shape the output.
Infosys, the nearly $19 billion annual software and consulting company that Nilekani co-founded in 1981, is positioning itself as an AI provider to its customers in greater than 50 countries, helping them organize their data and their very own models to coach.
Nilekani highlighted Infosys' company launch last month of two small language models in collaboration with AI chip company Nvidia. They were trained on Infosys data and integrated with products resembling digital banking software Finacle.
“We actually offer our customers a service to construct a model. . . There’s numerous interest in it because we’re demystifying all this modeling stuff,” he said. “All the technology has develop into so easy which you can construct models in a matter of months.”
Some analysts have expressed concerns that India's outsourcing industry will suffer as more firms adopt technologically sophisticated AI models, with value flowing to software providers and the so-called hyperscalers of cloud services. However, as firms develop their very own models, the technology will present a possibility, not a threat, to firms like Infosys.
Nilekani, a fatherly figure in India's tech industry who has played a central role in shaping the federal government's digital policy and the pioneering Aadhaar biometric identity system, said AI would replace some roles in his country's tech industry – but latest roles would emerge, who pool their power.
India's services sector has struggled with subdued growth and slowed hiring declines in recent times as clients including Wall Street banks cut IT spending.
Nilekani didn’t expect a pointy increase in headcount in India's technology services industry, which employs greater than 5 million people. Partly due to advances in AI, but in addition due to a weak global economy, they “may not grow as much as they’ve previously,” he said.
The industry is certainly one of India's few major sources of employees in a rustic of 1.4 billion those that is struggling to create meaningful jobs to soak up its young and enormous workforce.
At the identical time, the Infosys chairman said Donald Trump's re-election could boost the industry, at the same time as Indian IT firms were hit by his recent crackdown on H-1B visas that firms within the US use to recruit expert foreign employees for every week limited variety of years.
“The bullshit scenario is that this may trigger deregulation of the market, firms will grow, people will make more acquisitions and subsequently there shall be more activity and subsequently IT people will profit,” he said. “In my opinion they may probably support legal migration, which is sensible. I mean really top quality talent, why wouldn’t you?”
