While You’re Debating AI, Someone Else Is Getting Better at Your Job

By Joe “Crash” Kelley

For radio, AI is the assistant many of us could never get the budget for.

I’m one person handling programming, imaging, commercial production, music scheduling, and coordinating remote voicetrackers across five stations. There are only so many hours in a day, and only so much of one person to go around.

That is the reality I bring to the AI conversation. When the debate focuses on the jobs AI might eliminate, I think about the work already sitting on my desk—and the staffing that radio has already lost.

Federal employment data put some numbers behind that history. The Bureau of Labor Statistics’ radio broadcasting series shows 122,800 jobs in 2000 and 85,000 in 2019, a decline of about 31%. That substantial contraction happened before today’s generative AI boom. The figures do not explain every staffing decision, but they establish that radio’s employment problems did not begin with this technology. ^[1]

The work remained. Professionals still had to get the stations on the air, serve advertisers, connect with listeners, and make the product sound good. Any serious conversation about AI in broadcasting has to include the people carrying that workload.

I have used AI to automate repetitive tasks and strengthen skills outside my specialties. That gives me more time to apply the experience, judgment, and creativity I have spent a career developing. A few minutes recovered from one task can become a better commercial, a stronger piece of imaging, or the chance to give an idea another pass.

Research offers evidence that such gains are possible. In a randomized experiment involving 453 college-educated professionals, researchers Shakked Noy and Whitney Zhang found that access to ChatGPT reduced average completion time on professional writing tasks by 40% and increased assessed output quality by 18%. Those were specific assignments, not entire jobs, but the study demonstrated that speed and quality could improve together. ^[2]

For me, some of the greatest value has come through learning. AI has helped me become a better producer by giving me technical guidance I can question, test, and apply in my digital audio workstation. I have developed a processing chain that helps me achieve the kind of sound I once heard from major-market production talents and wondered how to create.

I write the words and melodies for my clients’ jingles. AI helps me develop those ideas into finished productions. My jingle and production work through SonicAttention now reaches 14 states, including a top-five market. I have more ways to turn the sound in my head into something a client and a listener can actually hear.

The idea of AI as a learning aid also has research behind it. A study of 5,172 customer-support agents found that AI assistance increased issues resolved per hour by 15% on average, with larger benefits for less experienced and lower-skilled workers. The researchers also found evidence of learning that persisted when the system was unavailable. The benefits varied; experienced, highly skilled workers did not gain in the same way. ^[3]

That distinction matters to someone who wears several hats. Decades of broadcasting experience do not make me a specialist in every technical or creative task that lands on my desk. I can be experienced in the profession and still have plenty to learn.

Another experiment, involving 791 Procter & Gamble professionals working on product-development challenges, found that individuals using AI matched the solution quality of two-person teams working without it. AI also helped participants produce proposals that combined commercial and technical perspectives. That suggests a practical use for professionals asked to work beyond their usual specialties. ^[4]

These studies did not examine radio production, and their percentages are not promises about what any station will achieve. They do, however, give us evidence worth considering alongside the experiences of people already using these tools.

My experience has also taught me to stay demanding. I still choose the idea, write and revise the copy, shape the melody, judge the performance, and decide whether the finished production is ready to air. I am accountable for what comes out of the speakers.

Research reinforces the need for that judgment. In an experiment with Boston Consulting Group professionals, AI improved speed and quality on tasks within its capabilities. On a separate task designed to fall outside those capabilities, AI users were less likely to reach the correct answer. Knowing when to question the output is part of knowing how to use the tool. ^[5]

There is a creative warning, too. An experiment involving short stories found that access to AI-generated ideas improved evaluations of individual stories, while making the stories more similar to one another. For a business built on distinctive voices and memorable brands, that is a reason to bring more of ourselves to the work: our ears, our taste, our local knowledge, and our willingness to reject something generic. ^[6]

I want AI to help me develop my ideas and make better decisions. That requires clear direction, experimentation, revision, and a professional standard for the result. Accepting the first thing a system produces is hardly a creative philosophy.

None of this makes endless staff reductions a good business strategy. Better tools do not erase the need for people, reasonable workloads, or investment in local broadcasting. We can challenge how companies use AI while recognizing what it makes possible for the professionals doing the work.

While so many remain caught in the argument over whether AI is good or bad, working professionals are using it to reach milestones that speak for themselves: learning skills, improving their work, and bringing ideas to life that once felt out of reach.

After decades in broadcasting, I still want to become better at this. A failure to keep learning can become a professional limitation of our own making. Experience gives us a foundation; continued learning determines what we can build on it.

Ignore it at your own peril. The tools will keep improving, and so will the professionals willing to learn them.

Research and source notes

1. U.S. Bureau of Labor Statistics, “Employment for Information: Radio Broadcasting (NAICS 51511) in the United States,” series IPUJN51511W200000000, via FRED. Annual values: 122.8 thousand jobs in 2000 and 85.0 thousand in 2019; calculated decline: 30.8%. This industry series counts jobs across occupations, including self-employment, rather than only on-air positions. Accessed October 6, 2026. Read source.
2. Shakked Noy and Whitney Zhang, “Experimental evidence on the productivity effects of generative artificial intelligence,” Science 381, no. 6654 (2023): 187–192. Randomized experiment using occupation-specific professional writing tasks. DOI: 10.1126/science.adh2586. Read source.
3. Erik Brynjolfsson, Danielle Li, and Lindsey Raymond, “Generative AI at Work,” The Quarterly Journal of Economics 140, no. 2 (2025): 889–942. Figures here follow the published article, rather than earlier working-paper versions. Evidence of lasting learning comes from performance during system outages; the authors note uncertainty in those estimates. DOI: 10.1093/qje/qjae044. Read source.
4. Fabrizio Dell’Acqua and coauthors, “The Cybernetic Teammate: A Field Experiment on Generative AI and Teamwork,” Organization Science 37, no. 4 (2026): 1217–1242. The published study analyzes 791 professionals. Its findings concern product-development tasks and do not establish that one person can replace a team across an entire job. DOI: 10.1287/orsc.2025.20702. Read source.
5. Fabrizio Dell’Acqua and coauthors, “Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality,” Organization Science 37, no. 2 (2026): 403–423. Field experiment involving 758 consultants, with results varying by the task’s suitability for AI assistance. DOI: 10.1287/orsc.2025.21838. Read source.
6. Anil R. Doshi and Oliver P. Hauser, “Generative AI enhances individual creativity but reduces the collective diversity of novel content,” Science Advances 10, no. 28 (2024): eadn5290. The experiment concerned short-story writing; the implications for radio branding are the author’s interpretation. DOI: 10.1126/sciadv.adn5290. Read source.