If you’ve spent any time reading about artificial intelligence over the past couple of years, you’ve probably seen countless articles suggesting that prompt engineering is the ticket to a high-paying career. The idea was simple: learn how to talk to AI models effectively and employers would line up to hire you.
A study published last month by GoHumanize, however, suggests that may no longer be the case.
The company analyzed 55 AI-related skills, comparing job demand with average salary data to determine which capabilities offer the strongest earning potential. While prompt engineering still appears to be a valuable skill, it did not come close to topping the list.
Instead, knowledge of large language models, often shortened to LLMs, emerged as the most valuable AI skill in the study. GoHumanize found nearly 57,000 active job listings seeking LLM expertise, with average salaries approaching $199,000 per year.
That finding shouldn’t be entirely surprising. Large language models power many of today’s most popular AI tools. As businesses rush to build AI-powered products and services, demand has grown for workers who understand how these models operate, how to fine-tune them, and how to deploy them responsibly.
Deep learning ranked second in the study. While it actually generated more job listings than LLM expertise, more than 67,000 according to the data, average compensation was lower at roughly $179,000 annually.
Computer vision came in third place, followed by AI Product Management and Natural Language Processing. The rankings suggest that employers continue to place a premium on specialized AI knowledge, particularly skills tied directly to building and managing modern AI systems.
Some of the highest-paying skills on the list had far fewer job openings. LLM fine-tuning, for example, posted an average salary above $208,000 but appeared in only about 7,200 job listings. Multi-agent frameworks and agentic AI skills also scored highly for compensation while remaining relatively niche compared to broader categories such as deep learning.
One of the more interesting findings involved AI Product Management. Unlike many of the technical skills on the list, product management focuses less on building models and more on deciding which AI products should be built, how they should be positioned, and how they can solve real business problems. According to the study, those responsibilities are commanding salaries that rival some of the most technical AI roles.
The study’s broader conclusion is likely to spark debate. GoHumanize argues that many of today’s most valuable AI skills can be learned outside traditional university programs through online courses, open-source projects, and hands-on experience.
While that may be true for some employers, it’s worth noting that many AI-related positions still prefer or require formal education, especially for research-heavy roles. Learning a skill and landing a six-figure job are not always the same thing.
The findings should also be viewed with some caution. GoHumanize is a company that sells AI-related software rather than an independent labor market research organization. While the data points are interesting, readers should keep the source in mind when evaluating the conclusions.
Still, the research offers an interesting snapshot of where AI hiring appears to be headed. Just a few years ago, prompt engineering was being promoted as one of the hottest opportunities in tech. In 2026, understanding the models themselves may be far more valuable than simply knowing how to write prompts for them.
Support independent tech journalism
NERDS.xyz is independently owned and operated. If you enjoy my coverage of Linux, AI, hardware, cybersecurity, and tech culture, consider supporting the site on Ko-fi.
Support NERDS.xyz