Tech Explained: Here’s a simplified explanation of the latest technology update around Tech Explained: India must speed up AI upskilling: Coursera cofounder Andrew Ng in Simple Termsand what it means for users..
The AI scientist, who earlier led Google Brain and served as chief scientist at Baidu, also cautioned against the swelling hype over artificial general intelligence (AGI), arguing that large language models are not a path to human-level intelligence and that claims of AGI being imminent are misleading.
“Today’s models have consumed more data than any human ever will, yet they’re still dumber than humans in many ways,” Ng told Sruthijith KK in a chat at ET House at the World Economic Forum on Monday. Edited excerpts:
What does AI mean for India and specifically for one of India’s biggest industries, which is the IT services industry. The rapid advancements in AI mean a lot of software coding and software development is done by machines themselves.
I think the worries about disruption are well placed. The risk and the opportunity is that with AI tools, individuals who are not skilled with AI are becoming dramatically less productive than people who are. Frankly, today, I would not hire a software engineer who does not know how to use AI tools in a fairly sophisticated way.
What is not well understood, but is spreading, is that for many other job roles from accountants to marketers, HR professionals to IT backend management, companies are demanding sophisticated use of AI skills, maybe even to the point of coding. So frankly, today, when I hire a marketer or HR professional, I would strongly prefer to hire someone who knows how to use AI to the level of being able to write software or build software to help be a better marketer or HR professional.
So for India, with its very large professional services and outsourcing industry, the risk and the opportunity is this: if this group of very smart, hard-working people, whom the world has trusted for a long time to do important work, can upskill rapidly, then India will continue to provide tremendous value. But if the upskilling doesn’t take place rapidly enough, then I think job disruption and job displacement could be significant.
So I feel a lot of pressure for India and a handful of other nations around the world to figure out the training and make sure we get this one right.
Since the ChatGPT moment in 2022, when you look at the evolution of these models and what they’re capable of, have you been really impressed, or have you been underwhelmed?
I think it’s been great. Since the first version of ChatGPT, the models have continued to improve. I feel like the hype has gone incredibly fast. The business reality has also gone fast, but not as fast as the hype, as things often are.
More work lies ahead of us to build AI applications, AI agents, and agentic workflows on top of this technology. The hype that we’ll get AGI (artificial general intelligence) in two quarters or a year or whatever, that’s just hype. I seriously doubt that’s going to happen for any reasonable definition of AGI.
But at the same time, without having computers that are as intelligent as humans in a general sense, there’s still so much work to be done in business process automation and re-engineering. I think there will be many big industry transformations across sectors, and we’ll still be working on this a decade from now.
Yann LeCun has recently said that he does not believe LLM-based models are the right way to even approach AGI, and that a totally different wave is required. Where do you stand on that debate?
I think he’s right. At the same time, the most reasonable definition of AGI I know is AI that can do any intellectual task that a human can.
So if we have AGI, it means AI should be able to spend five years writing a PhD thesis or learn to drive a truck through a jungle. LLMs by themselves are not a path to AGI. But there’s no technology I know of today that by itself is a path to AGI.
To achieve that type of AGI will need new technical breakthroughs that none of us really know yet. You can plug in any existing technology and say it’s not a path to AGI and that would be a true statement.
One missing thing is that because of the hype around AGI, I’ve seen businesses attempt to redefine AGI at a lower bar to make it easier to hit. This is problematic, because most of society thinks AGI means AI that is generally as smart as humans.
When companies redefine AGI and then say, under some narrow definition, they’ve achieved it, it misleads the public and business leaders into thinking AI is on a trajectory to be more powerful than it actually is. I’ve seen this mislead high school students making career decisions, and CEOs making very large capital allocation decisions based on assumptions I don’t think will materialise.
AGI has been defined and redefined so many times that it’s almost lost meaning. That said, even without reaching this really high bar of AGI, AI is incredibly exciting and incredibly valuable. India, the US, and other nations building toward that will be exciting.
I’m sure CEOs ask you all the time what’s happening and how they should respond. What’s the most common question you get?
My business partner Kirsty Tan and I do a lot of work through AI Aspire advising large businesses. The biggest thing we talk to CEOs about is how to help businesses adopt AI. The number one issue is people change.
How do we upskill people and take them along? How do we manage the people side of change so that the people who ultimately do the work can adopt AI?
Many businesses have tried a “let a thousand flowers bloom” strategy. Bottom-up innovation is important, but mostly this approach hasn’t worked. It often leads to point solutions, where a small piece of a process becomes much more efficient, but overall business gains remain incremental.
The bigger wins come from re-engineering entire workflows, often requiring top-down leadership or someone with a broader view. Where businesses get this right, they deliver much more value.
You spoke about hype running ahead of business reality. There’s a debate today between going all-in on AI and being cautious because tangible outcomes aren’t always visible. What’s the right approach?
Business outcomes can be delivered if you have people with the right skills to think through business processes and projects. If teams don’t understand AI well enough to select and implement the right projects, you won’t see results.
AI can make things much faster or much cheaper. But the real gains come when businesses use that speed to offer dramatically better products or services. For example, in loan underwriting, using AI just to save costs gives incremental gains. Using it to give customers a decision in 10 minutes instead of a week creates growth.
Leadership matters. My challenge to CEOs is this: if you’re serious about AI, learn it yourself. Spend a few hours understanding the technology so you can guide your teams, allocate resources, and drive results.
Is that a free course, or does one have to pay for it?
You can get all the videos for free. In all seriousness, I’ve seen that when CEOs and their top leadership teams take even a short course on generative AI, the sophistication of their decision-making changes dramatically.
I remember working with a company where leadership initially told my team to just work harder. After they took one of our online courses, the very next brainstorming session was completely different, much more informed, much more realistic. Just a handful of hours of learning can make a huge difference.
It seems to me that, unlike a lot of technology problems of the past, this doesn’t seem like something you can entirely outsource to consultants.
Some things should be built in-house, and for other things, finding the right partner makes sense. Businesses know their own domains best. Finance companies understand finance, healthcare organisations understand healthcare. They don’t need me to solve those problems. But they may need partners who are cutting-edge in AI to accelerate decisions.
Fundamentally, large language models predict the next word. How do they sound so intelligent if that’s all they’re doing?
Pre-training teaches the model to predict the next word using massive amounts of internet data. Then a second step aligns the model to be helpful, honest, and harmless, often using reinforcement learning. That second step makes a big difference in how useful the model feels.
This scaling approach has worked remarkably well. We knew this over 15 years ago when I started the Google Brain project. We’re still squeezing more performance out of this approach.
Do you think AI will advance pure science, solve long-standing mathematical problems, for example?
AI is already contributing to science, from biology to materials science. Truly original breakthroughs will come, but concepts like creativity and originality are philosophical rather than scientific questions. If AI behaves in ways we perceive as creative, I’m comfortable calling it creative.0
Is AI effective because it learns the way humans do?
AI learns very differently from humans. Today’s models have consumed more data than any human ever will, yet they’re still dumber than humans in many ways. Neuroscience has offered limited inspiration so far because we still don’t fully understand the human brain.
Did you imagine AI would reach this scale 15 years ago?
I’ve always been bullish on AI. For insiders, progress feels incremental. For the outside world, it looks exponential. ChatGPT felt sudden to many, but the building blocks were visible for years.
Which recent models have genuinely impressed you?
All of them. Claude, Gemini, OpenAI models, they’re all very strong. It’s great to see competition.
Many countries are talking about sovereign AI. How should India think about this?
Open-source and open-weight models are a powerful strategy. You don’t need to own everything, you need to ensure no one else controls your fate. Open models reduce geopolitical risk and help nations retain autonomy.
The US appears less welcoming to immigrants. How do you view that?
I think it’s a huge unforced error. Immigrants have been critical to US innovation. Making it harder harms both the economy and people’s lives. It’s been personally painful to watch.
Finally, what should India’s AI strategy be?
AI disrupts old rules. This gives India a chance to leapfrog again. The key is rapid upskilling. Ignore the hype, learn real skills, and build something bold and new.