Zerodha chief technology officer Kailash Nadh said it’s unrealistic to claim that AGI can be achieved within two to five years. Speaking at Cypher 2024, India’s largest AI conference by AIM Media House, he added that the definition of AGI has been set by Western AI companies.
“It must be their definition. They must have some definition that they made up for what intelligence is, and it might be very intelligent-looking behavior or technology, but for anyone to say that it will happen in two years or five years, no way,” he remarked.
He further explained that AI is fundamentally based on physics and biology, and that people have been making such claims since the 50s and 60s.
“People were trying to build AI using analog technology. So AI research and hands-on AI modeling predates general-purpose computing, and AI was always five years away for many, many years. Today, we have amazing breakthroughs built on many decades’ worth of research. But to say that in two, five, or 10 years we will have AGIs, I don’t know where that’s coming from — probably business reasons, valuation reasons,” he added.
Nadh said that today, AI is both ‘a lot of reality and hype.’ “AI research and computational AI machine learning research have been ongoing for 50 to 60 years, and some of the earliest use cases and attempts to use such technologies were to make money in the stock markets,” he said, adding that there’s nothing new about this and that he is amused by why it is in the news.
Zerodha Bets on Open Source
Nadh believes that the open-source community has to play a huge role in the generative AI boom. “I think this muscle and exponential explosion of generative AI technology is entirely driven by the open source ecosystem of tools—the entire stack, deployment tools, development tools, all kinds of things, evaluation tools, the frameworks, developer frameworks, the entire developer experience, and engineering experience,” he said.
He said that many open-source models available in the market are surpassing proprietary models in performance, and Zerodha is actively exploring these alternatives. He also mentioned that the company is using LLMs to address specific challenges.
“We are solving very specific problems within the organisation related to customer support and other areas. Recently, some of the systems we launched online, with just one developer experimenting with them, have effectively replaced the mundane manual tasks that a team of about 100 people used to perform,” he said.
Nadh highlighted how open source is essential to Zerodha and the broader tech industry. “Open source is fundamental to Zerodha’s existence. Without it, there would be no Zerodha, no Indian startup ecosystem, and no global technological explosion over the last two decades. The Internet runs on open-source technologies, and the largest government systems utilise them as well,” he said.
He further announced that next week the company is launching a dedicated open-source fund to support critical open-source projects globally. He added that Zerodha will also be producing a lot of open-source projects.
Notably, ‘Free and Open Source Software’ (FOSS) has been the pillar on which Zerodha is built. The company heavily relies on open-source software for its operations. Zerodha’s tech team funds and contributes to various FOSS projects. CTO Nadh has also developed Listmonk, an open-source newsletter and mailing list manager, which Zerodha uses to send millions of emails per month at a very low cost.
Will AI Surpass Human Intelligence?
Nadh said that Zerodha is one of the few companies that ensures there will be no job losses due to AI or any other technological advancements. “Last year, during the peak of the LLM hype, we made a decision to implement a policy explicitly stating that no one at Zerodha would lose their job solely due to the adoption of a particular technology.
“Instead, we would provide avenues for employees to migrate to other roles,” he said, emphasising a human-centric approach to technology.
He explained that the goalpost in AI development keeps shifting, with LLMs no longer being considered intelligent due to their widespread adoption. “All it took was 24 months for this shift,” he added, pointing out the speed at which expectations around AI have evolved.
“Five years ago, if somebody said that there’d be a chatbot capable of interacting exactly like a human being, solving real problems, and speaking in natural language, nobody would have believed it. But today, it’s become a commodity,” Nadh added.
Despite the rapid progress, Nadh believes that true artificial intelligence is still far off. “Will we eventually attain non-biological artificial intelligence? It’s highly likely, but it could be centuries from now,” he concluded.
The post ‘You Cannot Achieve AGI in Two to Five Years,’ Says Zerodha CTO appeared first on AIM.
