It’s a confusing time as the tech industry adapts to AI, and it’s hard to know what’s actually happening. Take Microsoft: A few days ago, news broke that the company was firing 6,000 employees. At first, a spokesperson told Alex Halverson of the Seattle Times that, “the layoffs weren’t performance-based but rather part of an overall strategy to streamline the corporate ranks and flatten management across the company.” In other words: “You all did great, and we’re the most valuable company in the world, but I have some bad news.”

Was this because of AI? That’s the assumption some people are making. The same article points out that relatively few of the 6,000 people were actually managers, and that “employees describe an increased emphasis on AI in their daily work…with more pressure to incorporate AI into their role.” 

Meanwhile, Walter Frick in Bloomberg drew the same connection, and quoted a researcher at the Brookings Institution, Molly Kinder, who said, “Microsoft’s layoffs in software engineering and AI roles are a bellwether for how generative AI is beginning to reshape work. … Coding is at the leading edge of AI integration, so it’s no surprise that engineers are among the first to feel the impact. They’re the first—but won’t be the last.”

That seems ominous. But then Satya Nadella, the CEO of Microsoft, is publicly stating that AI is going to drive more hiring, not less. In a Bloomberg Businessweek profile by Austin Carr and Dina Bass, which was mostly about the AI efforts at Microsoft, he repeatedly pushed a message of AI opportunity: “If Microsoft spends hundreds of billions of dollars on AI projects, [Nadella] estimates that will likely generate trillions of dollars in economic activity—more data centers mean more construction, energy use, manufacturing equipment and so forth.”

Backing that up is an article by Reed Albergotti in Semafor, where Nadella and his CTO, Kevin Scott, came down even more in favor of humans: Microsoft “plans to hire more software engineers than it has today, but it cares more about what makes them human and less about their technical abilities,” and Scott said he wants to hire people who are into “the full breadth of human history and understanding how social systems work.” (Emphasis added.) That’s quite the job description. I look forward to the day Microsoft appoints its Chief Weltanschauung Officer.

The same back-and-forth is happening across the industry. The CEOs of Shopify and DuoLingo have been proclaiming that AI will be core to their mission evermore—and that hiring humans is no longer a priority. But the CEO of Klarna is pulling back on their “no more hiring, just AI” commitment, finding that you need people after all. (Except this time they’d prefer to hire them part-time into gig roles, not full-time, which seems to be under-reported.) 

Meanwhile, economists at the University of Chicago and the University of Copenhagen published a paper, described in a Fortune Magazine article by Irina Ivanova, which found that “on average, users of AI at work had a time savings of 3%.” Which…isn’t very much. Especially when you consider that the official Danish work week is 37 hours.

So which is it? Is AI replacing programmers or not? Will the technology industry suddenly find itself in desperate need of humanists, to better steer the bots? Should coders learn to write poems? Is sociology the new product management? Is AI really only saving us 3% of time in the workplace? What’s really going on?

I’m going to give a really unsatisfying answer, but I hope it’s helpful even so. I think the answer is: All of the above. One of the things you learn as you deal with big structures is that they contain multitudes. Anything of a certain scale is no longer one thing, or two things, but rather a huge overlapping set of things. This applies to the New York Times, the Catholic Church, Facebook, and the United States of America. Every single one of these social constructs has vast bureaucracies, ardent defenders, brutal critics, and vast ranges in between. I could argue for and against each one all day long—and it won’t make any difference.

As we’re trying to rope the next iteration of our product over the line—come to the party on June 3rd! See the thrilling, inexpensive future of wonderful, boring business software!—we’re facing all the classic challenges of software development, plus all the challenges of AI-driven software development. That was exciting for a while, but as deadlines approach, it feels very familiar. AI isn’t replacing humans when it comes to getting things over the line; in fact, our engineering ranks are growing. And I don’t envy our engineering and design teams navigating all this.

You can tell me just about anything about AI right now—it’s taking jobs? Probably. It’s creating them? Sure. Killing the commons? Yes. Creating new forms of human connection? Why not? Ruining the environment? Nowhere near air travel, but also not great. 

But let me speak as my true self for a minute, as a deep nerd who loves process: The absolute most exciting thing that could happen in the industry right now would be for Microsoft Press to update Code Complete in a third, AI-focused edition and talk through how to write and manage software projects with these new technologies. Like a big, meaty, boring book with every detail. This would be more exciting than a new version of Claude, or faster video production. I’d buy 10 copies on paper and send them to the team.

I’m done with disruption. I want process, repeatable results, and efficiency. Let’s say it is just 3% efficiency gains—that’s not nothing! Could we get it to 5%? 10%? Coding can definitely be faster with AI, but, again, the last mile remains long. Can we talk about that last mile instead of waiting for a miracle? Maybe we need a whole new kind of “Agile Methodology” with vicious scope cutting and more focus on real-time, group interaction with chatbots. Or something! Give me a process!

Did Microsoft lay off 6,000 good people because of AI? The simplest answer is it laid them off because it’s Microsoft, and it felt it can make more money by doing so. AI is obviously a part of everything they do now, so I’m sure that’s a factor, but there’s no way it was the only one. 

Coincidentally, the 6,000 represent about 3% of the total MSFT workforce—so perhaps the layoffs do represent those 3% AI-driven efficiency gains after all? Regardless, I hope those people land well and find good jobs. I’m betting they’ll be fine, at least after a while, because there’s a ton of work to do.