AI's success must be measured in economic impact, not benchmarks: Satya Nadella

Time of India

17 hours ago

The true metric of success of artificial intelligence (AI) must be measured in terms of its economic impact and real growth in global gross domestic product (GDP), said
Satya Nadella
, chief executive,
Microsoft
.
'…the ultimate measure for me always is not some AI benchmark, it is the GDP growth in the real world that means it's the growth in a firm. And that's, I think, where we're seeing absolutely real evidence that this is scaling,' he said in a conversation with Ali Ghodsi, founder and chief executive,
Databricks
, at the company's annual conference in San Francisco on Thursday.
When asked about drawing the line between AI hype versus real value add, Nadella outlined how AI should serve as an amplifier of existing enterprise investments.
'If the new technology somehow builds on the investments they've already made and compounds the effects for them, then that is a true value-add.'
Nadella reflected on the acceleration in AI development across pre-training, inference, and application maturity over the last two years.
'It's the compounding effects of all the S-curves, right? There is the pre-training S-curve, and then there is now the test time compute S-curve. There is just the application tier getting a more mature S-curve… It's that combination that's been pretty stunning to see,' he said.
Discover the stories of your interest
Blockchain
5 Stories
Cyber-safety
7 Stories
Fintech
9 Stories
E-comm
9 Stories
ML
8 Stories
Edtech
6 Stories
He also described how human-AI interaction is progressing beyond conversational AI toward operational automation.
'We started by saying, wow, I can now have intelligence that I can chat with. Then we said, oh, we can assign tasks to the intelligence, and now we effectively can have digital coworkers.'
Nadella emphasised the enterprise-wide usage of AI agents along with tools like
GitHub Copilot
and Databricks connectors.
'You are now getting to a place where it's become standard issue… every employee inside the enterprise is able to either manage agents, create agents, orchestrate agents in order to create that next level of automation around them.'
The two companies announced a multi-year extension of their decade-long partnership for the
Azure Databricks cloud platform
.
Parallelly, Databricks also announced a new product partnership with
Google Cloud
to make the latest
Gemini
models available within Databricks' platform to build, deploy, and scale AI agents.
(The reporter was in San Francisco at the invitation of Databricks).