DEV finds 'AI' means tools

“AI experience” has turned into one of those phrases that sounds more technical than it often is. In a lot of job listings now, it does not mean “go train a frontier model” or “publish ML papers.” It means something much more practical — can you use AI tools inside real work, wire them into software, and make a team faster? That is the point Charlie Morrison makes in a DEV post from April 28, and it lines up with the broader hiring pattern he describes across his recent job-market writing. (dev.to) ### So what do employers mean by “AI”? Basically, they mean applied tool use. Morrison’s breakdown is blunt: when postings say “AI/ML experience preferred,” they usually are not asking for a research background. They are asking whether you can integrate an LLM API into a product, build a retrieval pipeline, and judge when AI is actually useful instead of forcing it everywhere. (dev.to) ### Why is that different from classic AI hiring? Because classic AI hiring was narrower. It pointed at machine learning engineers, data scientists, and researchers — people working on models themselves. The newer version is closer to product engineering with AI components attached. You are not inventing the engine. You are figuring out where to bolt it into the car without breaking the brakes. That is a different job, and a much bigger slice of the market. (dev.to) ### What skills actually show up? The interesting part is how ordinary the surrounding stack looks. Morrison says the baseline is still Python or TypeScript, APIs, system design, Git workflows, and CI/CD. Then the differentiators kick in — cloud infrastructure, containers, data pipelines, security basics, then LLM APIs, RAG systems, vector databases, (dev.to). (dev.to) ### Why are prompts and automation suddenly job skills? Because companies are buying outcomes, not vocabulary. If a marketer can use AI to turn one brief into ten campaign variants, or a developer can automate support triage, that creates immediate value. Hiring managers seem to be using “AI” as shorthand for “person who can improve a workflow with mo(dev.to)er-tech language. That last step is an inference, but it fits the skills Morrison highlights. (dev.to) ### Does this mean deep ML expertise matters less? For most openings, yes — or at least it matters less than people think. Morrison explicitly contrasts what companies are not usually asking for — PhDs, training models from scratch, research papers — with what they do want, which is working product experience. The strongest signal he names is not a credential. It is a GitHub project that solves a real problem. (dev.to) ### Why is this showing up now? Because the broader developer market is fragmenting. In Morrison’s April 29 post, he says software job postings are up 15% since mid-2025 even while layoffs continue, and the hiring growth is concentrated in specialized and AI-adjacent roles. Companies are still hiring, but they are hiring for narrower needs — integrati(dev.to) more like one layer in a larger delivery stack. (dev.to) ### Who should care about this wording shift? Job seekers first. If you read “AI required” and assume you need advanced ML credentials, you may screen yourself out for no reason. Hiring managers should care too, because vague AI language can attract the wrong candidates. The clearer version is something like: experience using LLM APIs, automating workflows, evaluating AI features, or shipping AI-enabled products. That is much closer to the work. (dev.to) ### Bottom line The phrase “AI experience” is getting demystified. In a lot of real-world hiring, it now means tool fluency, workflow design, and product execution. Not “can you build the model,” but “can you make the model useful.” (dev.to)