Or did it? Just as video never quite buried radio, AI may be engineering a rebirth — not an obituary — for the developers bold enough to adapt.
The Panic
In 1979, the Buggles declared it with synthesized certainty: Video killed the radio star. The song wasn’t just a pop hook, it was a eulogy for an entire class of talent. The voice that had commanded airwaves, the personality that needed only sound to hold a nation spellbound; all of it, supposedly, obsolete. MTV launched. The camera became the new microphone. An industry trembled.
Sound familiar? Replace “video” with “AI” and “radio star” with the job title of your choice, developer, analyst, copywriter, paralegal, radiologist, financial advisor and you have the dominant anxiety of our current moment. The fear is real. The disruption is measurable. But the eulogy, as it turns out, may be premature.
“The question was never whether video would change radio. It was whether radio had enough identity to survive the change and reinvent what radio even meant.”
“The question was never whether video would change radio. It was whether radio had enough identity to survive the change and reinvent what radio even meant.”
What actually happened to radio? It didn’t die. It evolved. Radio stars who adapted to the video era became multimedia personalities. Howard Stern moved to satellite. NPR deepened long-form storytelling. Today, podcasting, radio’s spiritual heir – is a $23 billion industry. The medium transformed. The practitioners who understood their own craft deeply enough to survive a format war? They didn’t just survive. They thrived.
Industry Trends: Beyond Software Development
Before we get to developers, it’s worth surveying the landscape. AI isn’t just knocking on one industry’s door. It’s rewriting the rules of production, creativity, analysis, and decision-making simultaneously across every knowledge-work domain. The Buggles’ video attacked one medium. AI is attacking the medium of thought itself.
Legal & Compliance
Contract review, discovery, and due diligence was work that consumed junior associates for decades now runs overnight. The associate role is being redesigned, not eliminated. The lawyers who thrive understand that AI is a research engine, not a judgment engine. Legal strategy, client trust, and courtroom presence remain stubbornly human.
Healthcare & Radiology
Diagnostic imaging AI outperforms human radiologists on specific benchmarks. Yet radiologists who use AI now catch more, faster, with more accuracy; expanding capacity rather than contracting headcount. The physician’s value shifted from pixel-reading to clinical synthesis and patient communication.
Content & Copywriting
First-draft generation is commoditized. The writers who survived are curators, brand strategists, and editors who shape AI output, not producers of raw volume. Voice, taste, and editorial judgment became the scarce resource.
Financial Analysis
Earnings summaries, portfolio screening, risk modeling all automated at scale. The analyst’s moat shifted to judgment, narrative, and client trust. Numbers got cheaper; insight got dearer. The advisor who can contextualize AI-generated analysis for a specific client situation is more valuable than ever.
Design & Creative
Generative image tools flooded the market with visual output. Designers who understand systems, brand strategy, and human experience became more valuable, not less. The ability to direct and evaluate AI output to have taste, turned out to be the skill that couldn’t be automated.
Software Development
AI Disruption Signal Strength by Developer Task Type
| Task | Disruption Level |
|---|---|
| Boilerplate Code Generation | ████████████ 93% |
| Unit Test Writing | ███████████ 87% |
| Bug Triage & Debugging | █████████ 74% |
| API Integration | ████████ 68% |
| System Architecture | ████ 29% |
| Domain Modeling | ███ 22% |
| Engineering Judgment | █ 11% |
Here is the uncomfortable truth the panic obscures: most software developers were never really paid to write code. They were paid to solve problems. Code was the medium — the magnetic tape, if you will. The real product was always working software that served a human need, embedded in an organizational context, constrained by real-world tradeoffs. AI doesn’t eliminate that problem. It changes the instrument.
Consider what GitHub Copilot, Cursor, and tools like Claude Code actually do in practice. They dramatically accelerate the translation of intent into implementation. A developer who understands what they’re building through the domain model, the distributed systems tradeoffs, the failure modes, the security surface; can now build it in a fraction of the time. The AI doesn’t replace the understanding. It rewards it with leverage.
The radio star who understood storytelling didn’t fear the camera. The developer who understands systems doesn’t fear the model.
But here is where the analogy holds its sharpest edge: the radio stars who got displaced were those whose only skill was reading someone else’s script into a microphone. The developers who face the greatest displacement today are those whose primary value was translating a spec into syntax without contributing architecture, judgment, or domain expertise to the process. That task, like the voice-over artist who could never become a personality, is being absorbed by the machine.
Radio didn’t die after MTV. It mutated. The medium compressed into a more portable, more intimate, more democratic form, the podcast, the satellite channel, the streaming audio platform. And in that mutation, it found new audiences and new economics. The stars who made it weren’t those who mourned the old format. They were those who seized the new one.
We are watching the same arc unfold in software development. The developer who designs multi-agent AI systems, who understands how to decompose complex workflows, define trust boundaries between autonomous agents, orchestrate tool-calling chains, and validate emergent outputs, is not a diminished version of yesterday’s coder. They are a more powerful, more leveraged, more indispensable professional. They are the podcast hosts of the software world.
“Every wave of automation creates a new class of professionals whose job is to direct, evaluate, and extend the automation. That class is always more powerful than the one it replaces.”
“Every wave of automation creates a new class of professionals whose job is to direct, evaluate, and extend the automation. That class is always more powerful than the one it replaces.”
Look at what’s actually happening in enterprise engineering teams right now. The organizations that are winning aren’t those that replaced their developers with AI. They’re the ones that repositioned their best developers as AI system designers — people who define agent behaviors, maintain context architectures, evaluate model outputs against domain-specific quality criteria, and translate organizational complexity into intelligent system specifications. The title may not exist yet. The work absolutely does.
Within the payment systems domain alone, where considerable time has been spent modeling capability taxonomies across ACH, Wires, RTP, FedNow, and ISO 20022 compliance, the developer who understands the domain deeply can now build in weeks what previously took months. Not because they’re smarter, but Because they’re directing a model that can translate their understanding into working systems at unprecedented speed. The domain knowledge doesn’t become less valuable. It becomes the primary bottleneck and therefore more valuable than ever.
The role doesn’t disappear. It transforms. Here’s what that looks like in practice:
| The Old Way | The New Future |
|---|---|
| Writes implementation code manually | Directs AI agents across systems |
| Debugs syntax and runtime errors | Evaluates model output quality |
| Translates specs into functions | Defines agent behavior and constraints |
| Reviews pull requests line by line | Reviews intent, not just syntax |
| Authors unit tests by hand | Designs test coverage strategies |
| Manages individual service logic | Orchestrates multi-service architectures |
| Documents what the code does | Specifies what the system should do |
| Valued for throughput of output | Valued for quality of judgment |
The leverage ratio is the critical insight. A developer who effectively directs AI tooling doesn’t produce 10% more output — they produce an order of magnitude more. This has happened before in every domain where powerful tools arrived. The mechanical press didn’t eliminate the writer; it made the best writers vastly more influential. The spreadsheet didn’t eliminate the financial analyst; it made the analysts who mastered it capable of producing work that entire departments previously couldn’t. AI code generation doesn’t eliminate the architect. It makes the architect’s decisions land faster and with higher fidelity than at any point in the history of software.
The Buggles were right that something ended in 1981. But they were wrong about what survived. Radio didn’t die. It shed the skin of a single format and found new life across dozens of others. The voice mattered, the story mattered. The relationship between a personality and an audience mattered. None of that was killed by video.
Artificaial Intelligence will not kill the developer. It will kill the developer who confused writing code with understanding systems. The one who survives, who thrives, is the engineer who knows why a system needs to exist, what it needs to do, how it needs to behave when things go wrong, and how to work with increasingly powerful tools to build it faster than ever before. That developer is not a radio star in decline. That developer is the podcast host with a global audience, a direct line to listeners, and a cost structure that would have seemed miraculous a decade ago.
The signal didn’t stop. It just changed frequency. Find the new channel. Tune in.
