The AI jobs boom isn’t unfolding quite the way many people predicted.
When ChatGPT arrived in 2022, we were told to expect a wave of entirely new careers, from prompt engineers to AI ethicists. Four years later, employers certainly want AI skills, but the roles they are hiring for tell a more interesting story.
Rather than creating armies of people with shiny new AI job titles, businesses are increasingly looking for marketers, engineers, project managers, analysts and other professionals who know how to use AI effectively within their existing roles.
So which AI skills actually matter to employers in 2026?
To find out, I looked at five of the most talked-about AI skills and compared the hype with what is happening in the job market. Some have exploded in demand, some are becoming baseline requirements, and others have barely registered with employers at all.
Prompt Engineering
This was widely expected to be one of the most in-demand skills of the generative AI age. Many predicted it could become a common job title as more companies began deploying chatbots and LLMs for numerous tasks. That isn’t quite the way things have worked out (yet), but prompt engineering as a skill is already highly in demand.
Statistics gathered by Stanford for their 2026 AI index suggest that 22,227 US job listings sought applicants with this proficiency in 2025, a huge increase from 6,152 the previous year. So while companies aren’t specifically hiring to fill prompt engineer roles, they are hiring marketers, analysts, project managers and engineers with strong prompt engineering skills to help them deploy AI across their organizations.
AI Ethics And Governance
This might be the skill showing the biggest gap between expectation and reality, which should probably be a cause for concern.
It could be that many businesses don’t feel that their AI strategies are mature enough to warrant hiring people with skills in this field. But according to the AI index, just 0.05 percent of US job listings specifically sought candidates with these abilities. This makes it the smallest of the 10 AI skill clusters tracked by Stanford’s report.
Is this a sign that industry isn’t yet treating oversight as an operational priority, preferring to see it as a problem to be dealt with when things go wrong? With the need to navigate increasingly complex and onerous regulatory frameworks (the EU AI Act, for example), this could be a mistake, and a lot of companies could be setting themselves up to learn this the hard way.
Agentic AI
This was the subset of AI skills that Stanford’s researchers found had increased by the largest amount, from just 151 mentions in 2024 job ads to 16,541 in 2025. Ads seeking professionals skilled in LangGraph, an open-source platform for deploying AI agents, increased by over 2,000 percent, from 194 to 4,294, in the same period.
Of course, not many people were even talking about AI agents back in 2024. So some of that fast growth is just due to the fact that companies are starting from scratch when it comes to hiring for agentic skills. Orchestrating teams of agents to work cooperatively, agentic specialisms like web browsing and online shopping, and UX design focusing on creating tools that can be used by both agents and humans, all fall into this category.
Machine Learning
Machine learning is best thought of as the technical nuts and bolts that make every application we call “AI” work. And according to Stanford’s research, it’s still the most frequently requested AI skill in job ads. As a technical skill, requiring knowledge of computer programming, mathematics and statistics, it’s the “hard skill” underpinning everything else on the list, even if we’re more used to hearing about LLMs or agents these days.
While these newer and more trendy terms cover technologies that handle human-machine interfacing and workflow automation, ML algorithms quietly power recommendation engines, predictive maintenance, fraud detection, risk modeling and all the vital workloads industry entrusts to AI in 2026. And they’re all built and maintained by humans with ML skills.
Chatbots And Conversational AI
Considering this is most people’s introduction to AI, and the application of AI that many businesses immediately recognize as valuable, it isn’t surprising they’re still looking for humans skilled in designing, deploying or just working alongside chatbots.
While the number of job ads requesting skills in this field continued to climb in the last year, from 5,535 to 14,376 for ChatGPT, they fell in proportion to the total number of AI-related job ads; in the case of “chatbot” by 50 percent, and “conversational AI” by 68 percent. So employers are still looking for people with these skills, but may be more focused on other fields such as agents or ML engineering right now. This doesn’t mean that skills in this field aren’t important anymore—chatbots are still fundamental to the way we use and interact with all forms of AI. So perhaps it’s just become a baseline skill we assume anyone interested in working in AI will have? Just like “proficient in Microsoft Word and Excel” would have been included in every CV 20 years ago, but not so much today.
AI Skills: The Big Picture
What stands out here is that AI isn’t necessarily creating new job titles yet, as was predicted might happen. Instead, AI skills are being absorbed into existing jobs and becoming part of the skill sets that every professional is expected to have.
Project managers are more likely to be expected to have an understanding of AI workflow design, software engineers will need to know how to use AI coding tools, and marketers will need to be comfortable with briefing and supervising genAI content tools.
This means that those with the most marketable AI skillsets in 2026 are those who can use AI to augment their own abilities, whatever field they’re in.
