Intelligence is foundation
Podcast Subscribe
Voices & Thought Leaders Saturday, 7 March 2026

Why Software Engineering Might Be the Last Job Standing

Share: LinkedIn
Why Software Engineering Might Be the Last Job Standing

Here's an uncomfortable idea: the people building AI automation tools might be the last ones with jobs. Not because they're special, but because of structural logic.

Latent Space's latest analysis argues that software engineering - specifically AI engineering - is the final profession to automate. Not the first. The last. And the reasoning is simpler than you'd expect.

If you're building the tools that automate other fields, you're always one step ahead of automation itself. Radiologists get replaced by diagnostic AI? Someone had to build that. Customer service gets automated? Engineers built the system. Legal research? Same story.

The people doing the automating are, by definition, the hardest to automate. They're building the thing that would replace them.

Jevons Paradox for Employment

The piece draws on Jevons Paradox - the idea that making something more efficient often increases total demand rather than reducing it. Coal engines got more efficient, so we used MORE coal, not less. AI makes coding faster... so we might need MORE engineers, not fewer.

Think of it like this: when spreadsheets automated accounting calculations, did we fire all the accountants? No. We gave them more complex work and hired more of them to handle the expanded scope.

If AI makes software development 10x faster, do businesses stop at their current software needs? Or do they expand into problems they couldn't afford to solve before? Probably the latter.

But here's where it gets messier. Jevons Paradox assumes demand is elastic - that there's always more work to unlock. For accounting, that held true. For software? Maybe. For other fields being automated? Less clear.

The Structural Advantage

Software engineers - particularly those working on AI systems - have a unique position. They understand how automation works. They can see what's coming. They can adapt their skills faster than most because they're building the tools themselves.

This doesn't make them immune. But it does give them time. And leverage. If you're writing the code that automates legal research, you're not just employable - you're essential. At least until the AI can write that code itself.

Which brings us to the obvious question: what happens when AI can build AI? When the loop closes? The analysis doesn't have a clean answer for that. Neither does anyone else.

What This Means for Everyone Else

If software engineering is structurally the last job, what does that mean for the rest of us? Should everyone learn to code?

Probably not. The logic here isn't "engineers are safe forever." It's "engineers are last in line." That's different. And it assumes a specific path of automation that might not play out as predicted.

More useful: understand that proximity to the tools of automation matters. The closer you are to building, deploying, or meaningfully directing automated systems, the longer your work stays relevant. The further you are from that, the faster your role gets compressed or eliminated.

This isn't about coding specifically. It's about being on the right side of the automation curve. Builders, for now, are on that side. But the curve is moving.

The Uncomfortable Bit

The real tension isn't whether AI engineers stay employed. It's whether the rest of the economy can absorb displaced workers fast enough. Jevons Paradox works when new demand emerges. But what if it doesn't? What if efficiency gains just concentrate wealth rather than creating new categories of work?

That's the question nobody wants to answer. Because if software engineering IS the last job, what happens when even that job is automated? Either we've solved scarcity and unlocked post-work abundance, or we've built a system that doesn't need most of us.

Right now, we're betting on the former. But we're not planning for the latter. And that gap - between optimism and preparation - is where things could get messy.

More Featured Insights

Builders & Makers
Why AI-Generated Code Becomes Impossible to Maintain
Robotics & Automation
Retail Robot Passes Safety Tests Designed for Public Spaces

Video Sources

Theo (t3.gg)
T3 Code launches as open-source agent orchestration platform
ArjanCodes
Stop Mixing FastAPI with Business Logic: Fix It with Ports & Adapters
Matthew Berman
The Future Live | 03.06.26 | Guests from Augment Code, NEAR Protocol, and Modular
AI Explained
What the New ChatGPT 5.4 Means for the World

Today's Sources

Towards Data Science
The Black Box Problem: Why AI-Generated Code Stops Being Maintainable
DEV.to AI
Building a Browser Game Center with Pure HTML/JS/CSS - Canvas 2D API and Game Loop Patterns
Hacker News Best
LLMs work best when the user defines their acceptance criteria first
DEV.to AI
How I Recovered 73 Accidentally Deleted Receipts Using the freee API
The Robot Report
Simbe Tally shelf-scanning robot achieves UL 3300 certification
The Robot Report
Plug-and-Play AI: Transforming robotics with modular skills
ROS Discourse
LSEP: Open protocol for standardized robot-to-human state communication
Robohub
Robot Talk Episode 147 - Miniature living robots, with Maria Guix
Latent Space
[AINews] AI Engineer will be the LAST job
Ben Thompson Stratechery
2026.10: Higher Powers and Lower Macs

About the Curator

Richard Bland
Richard Bland
Founder, Marbl Codes

27+ years in software development, curating the tech news that matters.

Subscribe RSS Feed
View Full Digest Today's Intelligence
Free Daily Briefing

Start Every Morning Smarter

Luma curates the most important AI, quantum, and tech developments into a 5-minute morning briefing. Free, daily, no spam.

  • 8:00 AM Morning digest ready to listen
  • 1:00 PM Afternoon edition catches what you missed
  • 8:00 PM Daily roundup lands in your inbox

We respect your inbox. Unsubscribe anytime. Privacy Policy

© 2026 MEM Digital Ltd t/a Marbl Codes
About Sources Podcast Audio Privacy Cookies Terms Thou Art That
RSS Feed