For Software Engineers, the AI Reckoning Has Already Arrived

Summary

---

Summary: Bloomberg Businessweek — For Software Engineers, the AI Reckoning Has Already Arrived

Date: July 17, 2026
Hosts: Carol Massar, Tim Stenovec
Guest: Mark Milligan (Deputy Editor, Bloomberg Businessweek)

---

Overview of the Episode

This episode dives into the transformative impact of AI on software engineers, focusing on how large language models (LLMs) like ChatGPT and Claude have rapidly altered the coding profession. Deputy editor Mark Milligan, who recently reported on this topic, joins Carol and Tim to discuss the emerging realities for tech workers: shifting job functions, new expectations within the talent pipeline, and existential questions about the future of programming. The conversation covers both the optimistic and unsettling dimensions of this "AI reckoning."

---

Key Discussion Points & Insights

1. The Disruption is Immediate and Profound

• Mark Milligan explains that AI's impact is not speculative: “This is like the biggest thing that's happened in their industry in at least 20 years.” (02:33, Mark Milligan)
• Software engineers now operate more as managers of AI models, frequently using tools like Claude to generate code instead of writing it themselves.
  • Quote: "Now they're not writing code anymore. They're just having conversations with Claude.” (02:58, Mark Milligan)

2. “Vibe Coding” and the Changing Nature of Work

• Carol jokes about “vibe coding,” asking if the term fits (03:24, Carol Massar).
• Mark classifies the shift as giving broad instructions to AI, which now “does in 10 minutes what used to take a person several days.” (04:08, Mark Milligan)
• Product management skills are emerging as a central requirement for engineers, as they orchestrate multiple AI “workers” on various features simultaneously.

3. Shifts in the Engineering Talent Pipeline

• Early-career layoffs are intensifying:
  • “Since ChatGPT came out in 2022, coding jobs for people in their 20s is down 20%.” (05:55, Mark Milligan)
  • Not purely due to AI—overhiring during the pandemic also plays a role, but AI is the underlying current driving new hiring and training philosophies.
• Even students are abandoning traditional CS majors for fields like cybersecurity, indicating acute anxiety and adaptation in university computer science programs.

4. The Elusive ROI and Productivity Gains

• Tim and Mark discuss how difficult it is to quantify AI-driven productivity:
  • One cited study found no measurable difference, but engineer sentiment is overwhelmingly pro-AI:
    • “They thought they were more productive, but they’re not actually more productive.” (07:22, Mark Milligan)
    • Yet since newer, stronger AI releases (Claude’s latest), engineers refuse to return to hand-coding for studies, suggesting qualitative gains that are hard to measure.

5. The Current Limits of AI Coding

• Mark highlights a few areas still resistant to full AI takeover:
  • Outdated hardware, niche platforms, or very specialized codebases
  • “Very niche types of electronics require a human touch... but they're still using Claude and Codex.” (08:11, Mark Milligan)
• Even in those scenarios, AI tools are central to workflow; engineers mostly step in to correct errors rather than design code from scratch.

---

Notable Quotes & Memorable Moments

• “Now they're not writing code anymore. They're just having conversations with Claude.”
  — Mark Milligan (02:58)
• “Here's how I want the app to be structured. Here's how I want...”
  — Mark Milligan, describing prompt-driven coding (03:48)
• “Since ChatGPT came out in 2022, coding jobs for people in their 20s is down 20%.”
  — Mark Milligan (05:55)
• “They would use Claude, they would find errors, they'd fix those errors themselves, but they're still using Claude and Codex.”
  — Mark Milligan on persistent human role (08:20)
• “It's almost shocking how quickly we are here.”
  — Tim Stenovec (08:35)

---

Timestamps for Key Segments

• 02:29 — Introduction to the disruption and Milligan’s reporting tour
• 03:26 — The concept and perception of "vibe coding"
• 04:20 — Boris Czerny at Anthropic and the new universal expectation for “a little coding”
• 05:30 — Impact of AI on the Silicon Valley job pipeline, students’ career pivots
• 06:51 — The challenge of measuring AI-driven productivity and ROI
• 07:59 — Current technical limits of AI coding (niche cases)
• 08:35 — Reflection on the speed of change

---

Tone and Language

The discussion is energetic, insider, and occasionally incredulous—matching the mood of a tech sector in fast, uneasy transformation. Humor surfaces in exchanges about “vibe coding,” but the mood turns somber when addressing layoffs and early-career anxieties. Milligan’s reporting provides grounded, firsthand insights, while Tim and Carol try to contextualize these changes for a broader business audience.

---

Final Takeaways

This episode paints a vivid picture of software engineering in the AI era:

• Tools like Claude and ChatGPT have upended traditional coding roles, leading to more productivity but also widespread job insecurity, particularly among entry-level engineers.
• The tech industry, universities, and individual engineers are all scrambling to adapt, but the pace of change is “almost shocking.”
• While some legacy coding problems still need a human touch, the message from Silicon Valley is clear: the reckoning isn’t coming—it’s already here.

---

Transcript

[Fidelity Advisor]

As a top performing advisor, you demand top performing funds. That's why Fidelity's 1000 experienced research professionals combine deep insights and advanced analytics to deliver consistent outcomes for clients across market cycles. Explore 300 plus Morningstar rated 4 and 5 star funds at I.fidelity.com Topfunds across all Fundshare classes of Fidelity, Fidelity Advisor Shares and Fidelity ETFs as of 6, 15, 2026. Past performance is no guarantee of future results.

[Bank of America Representative]

Introducing B of A Rewards A new loyalty program with rewards for every ambition. From cash back deals on brands you know and love to a credit card rewards bonus. From fueling up to rewards that fuel your goals. It all starts with a Bank of America checking account and grows from there. What would you like the power to do? Bank of America Open or enroll your [e-mail] B of A rewards bank of America Corporation All Rights Reserve the thing

[IBM Representative]

about AI for business, it may not automatically fit the way your business works. At IBM, we've seen this firsthand. But by embedding AI across hr, IT and procurement processes, we've reduced costs by millions, slashed repetitive tasks, and freed thousands of hours for strategic work. Now we're helping companies get smarter by putting AI where it actually pays off, deep in the work that moves the business. Let's create Smart to Business IBM

[Tim Stanvak]

Bloomberg Audio Studios Podcasts Radio news.

[Carol Massar]

You're listening to Bloomberg Business Week with Carol Massar and Tim Stanvak on Bloomberg Radio.

[Tim Stanvak]

All right, so we kind of want to stick with AI and trying to get to. I mean it's part of almost every story that we get to. And a fascinating story. It's a most read on the Bloomberg it's by our businessweek team about how the reality of AI and its impact and AI reckoning, you might say, has already arrived for software engineers. We want to get to this story. Mark Million wrote it. He's Bloomberg businessweek Deputy editor, joining Tim and me right here in studio. You know, we talk about first of all, great story. Loved it. We caught our attention this morning on our read in. We talk about ROI and productivity gains. Coders, software engineers are already feeling the impact of what I can do for their world. Maybe there's an upside and maybe there's a downside. But walk us through what's going on.

[Mark Milligan]

Yeah, I used to live in San Francisco. I covered the tech industry for more than 15 years. I'm living here now working on BusinessWeek. But I had just been hearing for the last six months from sources and old friends out there that this is like the biggest thing that's happened in their industry in at least 20 years. And I felt like I just had to go out there and see and just really understand what was going on because you hear a lot of doomers talking about how AI is going to take all the jobs. And here they're saying that because it's taking their job or it's at least drastically changing their jobs. And so I went into Anthropic and OpenAI, spent time at the labs and just met with people who have been doing this their whole lives, their whole adult lives, learning to code coding. And now they're not writing code anymore. They're just having conversations with Claude.

[Carol Massar]

Meaning they're vibe coding or something different.

[Mark Milligan]

Well, vibe coding, because there's not necessarily

[Carol Massar]

a name for this.

[Mark Milligan]

Yeah, they look at it as a little of a derogatory term.

[Carol Massar]

So that's more elementary. That's for the folks like me who don't actually know how to code.

[Mark Milligan]

Exactly. So there have been different terms for most of them, just call it coding, even though they're not actually writing the code themselves. But yeah, it's like basically telling it what you would have spent the next six hours doing. It's like, here's how I want the app to be structured. Here's how I want.

[Carol Massar]

Okay, so then what do they spend the 6 hours doing?

[Mark Milligan]

Now they spend the 6 hours building 10 different features for their app. And so they'll, they'll open separate windows and saying, like, you work on this, you work on this, you work in that. It's sort of like a product manager with which is a job that exists, but, like, those people are. Don't have people who are, you know, going to produce things as quickly, you know, within 10 minutes. That would take a person, you know, a couple of days.

[Tim Stanvak]

Mark, tell us about Boris Czerny out in San Francisco. He created the software Cloud code, runs it at Anthropic. And in your reporting you say that at Anthropic, the expectation that everybody is at least a little bit of a coder has become endemic. And that mindset is beginning to infect much of the tech industry. Tell us a bit more about. I'm assuming you chatted with him, you talked with him.

[Mark Milligan]

Yeah, I spent a bunch of time with Boris. He's a really interesting guy. He's like a coders coder. And so he's like this, a little bit of a guru that, like, people follow him on their podcasts and his ex post just like, figure out, like, how is my job going to change tomorrow. Like Anthropic is releasing stuff almost every single day. But he has a long background in the industry. He was the leader of code quality at Meta. So he kind of set the standards for how Meta developed Facebook and Instagram and worked on the server architecture and Instagram. And he's just sort of articulated what this new world looks like and it's changing even for him every week. Instead of actually telling it what to do now he just like creates this broad framework for his AI and just like go spend the next like several days working on this challenge.

[Carol Massar]

What does this mean for the, the, like the, what we've seen with the, the talent pipeline in Silicon Valley over the last 20 or 30 years or so. It comes to mind I was, I was actually speaking with a Bloomberg intern. We have a great class of interns here. He's not on the media side. And I asked what he was studying. He said, well I, I now studying cybersecurity, but I spent my first year majoring in computer science and programming. And he saw the writing on the wall. He decided to switch his major as a result.

[Mark Milligan]

Yes, it's been a sea change within especially entry level jobs in this industry and it's still so new that it's hard to know how it's really going to shake out. There's plenty of optimists who say there will just be more demand than ever for software and people who understand like how software fits together. I talked to somebody at Carnegie Mellon who's working on helping restructure their computer science program. But the early signs have not been great. So since ChatGPT came out in 2022, coding jobs for people in their 20s is down 20%. And it's a lot of the first in, first out at the big companies. They're doing mass layoffs. Some of them are blaming AI when it's really maybe not AI's fault. And they just maybe over hired during COVID and are trying to correct that. But for a lot of companies they're, they're true believers that this is a new way of working.

[Tim Stanvak]

Measuring productivity or ROI though is tricky. Even among these folks who maybe are finding they can, you know, basically use AI and talk to it and say go for five or six hours and do something and then maybe do another one somewhere else. But it's still tricky.

[Mark Milligan]

It's true. Everything based on my reporting leads me to believe that this dramatically increases productivity for a software engineer. But finding actual data to support that has been very difficult. There is one study that was highly respected. That took place early last year, before this was like real, before AI coding was really a common thing and that the models weren't as good. And they found that, well, people thought they were more productive, but they're not actually more productive. But there was this moment in time last November when, when CLAUDE released a new model that really was capable of, of replacing parts of people's jobs and, and ChatGPT followed with a similar model, they couldn't run the experiment again because all the software, most of the software engineers I went to were like, we don't want to hand code anymore. Like, we got a better way. We're not going to waste our time for your experiment.

[Carol Massar]

What's the, what's the thing now that you saw in Silicon Valley that still can't be done by Claude, that people are having to do that maybe a year from now Claude will be able to do?

[Mark Milligan]

You know, there are some examples of like, people who are coding on very, like, outdated hardware that had severe limitations, that claude, which is trained on incredible amounts of code bases to understand the best practices. There are like very niche types of electronics that require a human touch. But for the most part, those people, they would use Claude, they would find errors, they'd fix those errors themselves, but they're still using Claude and Codex.

[Tim Stanvak]

It's almost shocking how quickly we are here. I know we've been talking about three years, but it does still feel like pretty shocking.

[Carol Massar]

Mark, glad we're still here right now.

[Tim Stanvak]

Sorry, Mark.

[Bank of America Representative]

Thank you.

[Tim Stanvak]

Great story, Mark. Millie in Bloomberg Businessweek Deputy editor. Check out his story. It's on the terminal.