Protecting sensitive client data with LLMs

Technology is reshaping the world at a pace few people, inside or outside the industry, expected. But every so often, you meet someone who has not only witnessed the major waves of technological change, but helped build them. In this conversation, Marcus Fontoura, Technical Fellow and Corporate Vice President at Microsoft, CTO for Azure Core, walks us through the story of AI, what leaders are getting wrong, and how to develop the one thing that will matter more than any model or algorithm: human agency.

Marcus has lived through every major inflection point: early search, the rise of cloud computing, and now large-scale AI systems.

One of the first things he challenges is the popular narrative that we are heading toward an AI apocalypse, or an AI utopia. Both extremes, he explains, miss the point:

"My approach was more like, let me just explain what the technology is and what it does… it's basically a prediction system."

Marcus offers a clear explanation of modern AI. He compares today's large models to a system that has:

"Read nonstop for fifty thousand years… with near perfect memory."

But this doesn't make AI a mastermind. It makes it a stochastic parrot, extraordinarily capable, but not self-directed.

He also emphasizes that while AI will automate the mechanical layers of work, it will amplify, not replace, the leaders who know how to think:

"If your job is typing in a spreadsheet… then I would feel scared. But if you have the knowledge and experience to really add value, I wouldn't feel scared."

His point is: the danger isn't AI. The danger is becoming someone who only performs tasks AI can do.

We also cover the uncomfortable but increasingly visible trend: people relying on AI so heavily that they lose their independent critical-thinking muscles. Marcus acknowledges the risk:

"That is a little bit concerning… we will see good uses of technology and uses we don't want to happen."

He stresses that organizations must raise the bar for juniors, not lower it, and that AI helps experts more than novices:

"More experienced folks already know what to expect… junior employees may not know what is correct or incorrect."

This is one of the most important insights in the entire episode: AI accelerates expertise; it does not create it.

On hallucinations, Marcus is exceptionally candid:

"The more we use it, the more you have techniques to avoid it… but we have to double-check those things."

On leaders fearing displacement:

"Use AI in a way that amplifies your skills… automate the mechanical tasks and focus on what only humans can do."

And on what truly matters in this moment of technological upheaval:

"Technology shouldn't influence us. We should influence what we want to see in our society."

And he gave a useful explanation of the names of ChatGPT models:

"When you say that bigger AI models, when you move from ChatGPT three to four, four to five, basically these models have more parameters. So this means that you read a lot more, but also you memorize a lot more."

This conversation is a reminder that the most important focus should not be AI, it's the leader using AI with judgment, clarity, and agency.

Get Marcus's book, Human Agency in a Digital World, here: https://shorturl.at/v0lo8

Claim your free gift:

Free gift #1 McKinsey & BCG winning resume www.FIRMSconsulting.com/resumePDF

Free gift #2 Breakthrough Decisions Guide with 25 AI Prompts www.FIRMSconsulting.com/decisions

Free gift #3 Five Reasons Why People Ignore Somebody www.FIRMSconsulting.com/owntheroom

Free gift #4 Access episode 1 from Build a Consulting Firm, Level 1 www.FIRMSconsulting.com/build

Free gift #5 The Overall Approach used in well-managed strategy studies www.FIRMSconsulting.com/OverallApproach

Free gift #6 Get a copy of Nine Leaders in Action, a book we co-authored with some of our clients: www.FIRMSconsulting.com/gift