“I would say – all up in the world – the inflation-adjusted, economic growth has been pretty weak and one of the things that I’m optimistic about is digital technology can help boost it, and things like artificial intelligence can help boost it,” he said.
“That said, in the tech industry, we grow multiple times GDP. I think we are going to go through a phase today where there is going to be some amount of normalisation of the demand.
“Quite frankly, we in the tech industry will also have to get efficient right it’s not about everyone else doing more with less we will have to do more with less, so we will have to show our own productivity gains with our own sort of technology.”
Nadella told Schwab generative artificial intelligence was the most profound change in technology platforms in 15 years.
“If you look at it (AI), it’s at the beginning of a new S curve and that’s the fun part of being in the tech industry where there are certain technologies that are reaching maturity that need to get deployed and show real results, and then we now have essentially the emergence of a completely new set of technology, which I think is going to be revolutionary.”
He said the last time the tech world had major platforms being born was in 2007 and 2008 when the mobile platform and cloud platforms emerged.
“In the last 15 years they’ve gone mainstream – we’re still in the throes of its penetration, but it’s significantly understood that this technology is making a real difference,” he said.
“I think that the AI piece in this particular generation of AI is showing that type of I’d say ‘platform shift’.
“But it turns out that even the ChatGPT moment has I think captured people’s imagination. But we started the work with OpenAI two and a half years ago when we started building the AI supercomputer in Azure to train these large models. (Azure is Microsoft’s cloud service).
“So, in fact, the workload of this particular form of AI requires a complete rethink, even in the system architecture of the computing infrastructure, and we did that hard work and trained these models and when you look at GPT3.5 compared to what’s coming these are nonlinear developments.
“They’re showing emergent capability. And I’m not saying this is the last model architecture, innovation, there will be more to come, but the fact is that these things by themselves are becoming platforms that I think truly can make a difference.”
Adoption speeding up
Nadella went on to give an example of how ChatGPT is being used by Andrej Karpathy, a computer scientist who worked as the director of artificial intelligence and Autopilot Vision at Tesla.
“On New Year’s Day, I saw this tweet by Andrej Karpathy, who was an ex-founder of the autopilot group at Tesla, who’s an AI developer,” Nadella said.
“He referenced perhaps the biggest product that made a difference in his life in the last 12 months was the GitHub Co-Pilot we had launched last year.
“So, he’s a software developer using tools and saying, now GitHub Co-Pilot is generating 80 per cent of his code. So, that doesn’t mean Andrej is not writing new code and being at his creative best.
“It just so happens that now he has 80 per cent leverage in doing what he’s doing. He’s still the pilot. He does have a co-pilot.”
The significance of these comments on the same day Microsoft sacked 10,000 employees cannot be understated.
Microsoft is not only leading the charge in adopted generative AI to make its operations more efficient, earlier this week it included the ChatGPT app in all of its cloud services.
In other words, Microsoft is supplying the largest companies in the world with the tools to do what it is doing, replace thousands of people with more efficient ways of doing business.
Nadella used a second anecdote from India to stress how quickly ChatGPT can spread. The software has been used to allow access to all Indian government services in different Indian languages. Nadella cited the example of a farmer in a remote village using the new service.
He said the fact that this was made available and is working across India within three months of it being released is a point of how fast this technology will be adopted around the world.
“That basically meant that a large model, or foundation model that was developed on the west coast of the United States a few months before had made its way to do a developer in India, who then added value to it to make a difference in remote village life, and I’ve never seen that type of diffusion.
“We’re still waiting for the industrial revolution to reach some large parts of the world 250 years later. The internet maybe took 30 years. Maybe the cloud and mobile took 15 years. And now I think we’re talking months.”
Disclosure: The author’s self-managed super fund has shares in Microsoft.