Between 2024 and 2034, the U.S. Bureau of Labor Statistics projects 34% growth for data scientists .
Presenting a generative AI program with information, a request or a problem to solve is called inputting a prompt. The better prompts you input, the better results you can expect to receive as output. But crafting a helpful prompt is more than simply telling a program to write a recipe using the ingredients in your refrigerator. Learning how to be good at engineering prompts involves a blend of human logic, communication, understanding patterns and bringing your personal knowledge to the table.
"Ultimately, AI is interdisciplinary," says Lance Cummings, an English professor at the University of North Carolina Wilmington who applies structured approaches from technical writing to creating prompts. "So to understand it, you have to understand it from different perspectives, and certainly to implement it. It's not just going to be computer scientists out there making it happen."
According to an April 2026 report from the National Association of Colleges and Employers, nearly a third of employers surveyed say they're looking for early-career employees who can use artificial intelligence in their work. Sixty percent report assigning AI-related tasks to interns.
"Finding people who have the title purely 'prompt engineer' I would say is a rarer thing, in my experience," says Jules White, a computer science professor at Vanderbilt University in Tennessee and senior adviser to the chancellor on generative AI in enterprise education. "You find lots of people who are effectively playing this role and using these skills, but they may have another name, like AI agent engineer or data scientist or data analyst."
The U.S. Bureau of Labor Statistics projects 34% growth - a numeric addition of 82,500 roles - for data scientists between 2024 and 2034. This eclipses the 3% projected growth for all occupations during the decade and reflects employers' desire to make data-driven decisions. But due to AI's applications in various disciplines, prompt engineering isn't a skill exclusively for data scientists.
"Everybody's going to have to be a prompt engineer to a degree, whether they like it or not, over the next 10 years," White says. "Most jobs these days are difficult if you can't read and write. It's going to be as fundamental over time as reading and writing, your ability and skills with this."
"To put it simply, prompt engineering is how people communicate with AI systems to get better, more accurate results," says Yeqing Kong, an assistant professor of technical communication at the Georgia Institute of Technology's School of Literature, Media and Communication.
Prompt engineering is not tinkering with keywords that will unlock a perfect response. Instead, it's about providing the AI platform with the proper context and information, as well as recognizing patterns and revising prompts, to get the desired output.
It's also typically a skill that employers look for as part of a larger skill set, rather than a role or job title itself.
Cummings prefers using the term "prompt design" instead of "prompt engineering."
"Because really what we're doing is designing natural language in ways that elicit appropriate responses from machines or AI," he explains.
Martin Jones, a professor of artificial intelligence, law and ethics at Anderson University in South Carolina, likens engineering a generative AI prompt to assigning work to a teaching assistant. Jones says that if he gave generic instructions such as, "I want you to help me conduct some research," he wouldn't expect a good response because more specific instructions were lacking.
"They wouldn't know," says Jones, who is also associate dean of the university's College of Business and Economics. "What research? Where should I go? Are there any limitations? What exactly do you want me to do? What do you not want me to do? The same is absolutely true with generative AI tools."
Getting effective results would require sharing the research topic, telling the teaching assistant where they might find relevant information, and anything related that shouldn't be included - for example, excluding articles published before 1990.
Jones suggests generative AI users "introduce" themselves to the bot as they train it; for example, "I'm a professor and I teach ethics" can help establish context for the requests the user might make.
White references someone he knows who has lengthy conversations with generative AI to establish not only what they're looking for from the output, but also how to determine when the output is desirable.
"He starts by, let's discuss the problem that we're going to solve and the fundamental characteristics of the problem we're trying to solve, how we might know if we've achieved the solution, and how we might go and benchmark the solution," he says. "And he spends a ton of time doing that before he ever tries to solve the problem."
While this type of thorough prework is a heavier lift upfront than going directly into inputting a question or command, it can lead to the AI bot producing more accurate results and possibly result in fewer revisions until the desired outcome is reached.
"In an organizational context, a prompt engineer helps organizations use generative AI more effectively," Kong says. Specifics vary by industry, but the root responsibility is creating a set of instructions that the AI tool can use to solve a problem or meet a goal.
The person or team who takes ownership of crafting and inputting prompts should ensure all stakeholders are on the same page when it comes to using generative AI. Employees should read and understand their employer's policy about AI usage. And if multiple people within an organization are using generative AI for the same or similar tasks, establishing best practices for designing prompts can help lead to more consistent results across teams.
Job titles that may include prompt engineering among their responsibilities include AI agent engineer, large language model engineer, computer systems analyst, technical solutions architect and other roles in programming or data science. But because this skill isn't exclusive to technical roles, professionals across various disciplines may find themselves engineering prompts to complete tasks at work.
Software engineers and computer science professionals who design AI software or use it to help write code must have coding experience, but this isn't necessarily the case for those who use AI platforms.
Regardless of the discipline in which they work, people who write effective prompts should possess the following skills:
Most people who perform prompt engineering or prompt designing as part of their professional duties don't have the job title "prompt engineer." But some websites report salary data on the ones that do. Salary data from Glassdoor says the pay range for prompt engineers across all industries and all experience levels is between $104,000 and $168,000 per year. ZipRecruiter reports that the average annual salary for prompt engineers in the U.S. is $62,977, substantially less than the range reported on Glassdoor.
The BLS reports that data scientists make an average annual salary of $126,800, and that software developers make $148,100 each year, per May 2025 data.
Instructors at the collegiate level can make six figures as well - the BLS says the average annual salary for postsecondary teachers is $103,470 - but that can vary widely depending on the subject taught, years of experience and whether they work at a public or private university.
For in-person roles, higher salaries typically correlate with a higher cost of living. They also may vary by industry, skills and job responsibilities.
While there's no official pathway to learning prompt engineering or design, the following steps can lay the groundwork for developing this skill.
It's not necessary to take a course about prompt engineering, but doing so could be helpful to understand how humans interact with artificial intelligence. Experimenting with platforms such as ChatGPT and Claude can help users familiarize themselves with each program's interface and how they operate. Those who work for organizations that use Microsoft or Google platforms should get to know Copilot or Gemini, respectively. Whatever platform individuals use and however they intend to incorporate AI into their workflows, learning the basics - including what large language models are and what these programs can and can't do - is a good place to start.
To really practice designing useful prompts, users should determine a practical task in their everyday life that AI might assist with. Individuals should consider their areas of expertise, field of employment and hobbies or interests. What do they want to bring to the interaction, and what do they want the generative AI to produce? A small business owner will likely use AI differently than a college professor, for example.
Cummings says he encourages his students to think of a workflow that's part of their day-to-day life, what the individual steps are and which step or steps artificial intelligence might help streamline.
"Develop a prompt for that," he says. "What would that mean to make that an automation? And that gets you thinking about how to solve specific problems."
Rather than just asking the chatbot random or casual questions, users should pay attention to the structure of how they phrase prompts. They should notice patterns in the way they and the bot interact, and keep notes on what works and what doesn't work. Developing structured approaches and applying computational thought can demystify the process and help users develop the framework that works best for their individual situation.
There is no secret ingredient to crafting the perfect prompt. As with other skills, becoming fluent in prompt engineering requires practice, changing things that don't work effectively and trying again and again until the desired output is achieved.
"A good prompt engineer has to be curious, precise, analytical and - most importantly - willing to revise, because it involves lots of trial and error," Kong says. "AI outputs are not always predictable and accurate. They need to be comfortable with that kind of iteration."
Of course, anyone using AI - or any digital platforms, for that matter - should learn about the platform's privacy policy and think carefully about what personal information they share.
"A caveat that comes with it is, you should really be aware of the tool that you're using and how it's using the data you are giving to it," Jones says. "All of our input is being used somehow by these bots. Data privacy is a legitimate concern."
User information could be used to train the AI bots, and individuals may adjust settings to prevent their data from being used for this purpose. Those who access an AI agent or platform on company equipment or through a company account should be aware of their employer's policies regarding responsible AI usage and data privacy.
Because the technology is constantly evolving, industry experts don't necessarily think prompt engineering alone is a marquee skill with staying power - after all, there are bots that can write prompts and coach human users how to improve prompts. But it's a foundational skill that users of generative AI need to know how to do. The ability to work efficiently with machines while using human discernment to solve problems is relevant for today's job market - and likely tomorrow's.
Citing a 2025 Wall Street Journal article, Kong says, "Even though prompt engineering was described as the hottest AI job of 2023, people worry that it has already become obsolete as models improve. So there is a lot of uncertainty involved in the future of prompt engineering."
While prompt engineering itself is disappearing as a specific role, the skills required to do it well - critical thinking, problem-solving and effective communication - are timeless and transferable. AI platforms are the latest place professionals can demonstrate these skills.
Jones acknowledges that some roles likely will be reduced, or require retraining, due to AI. But to other professionals who are concerned about AI's effect on the job market, he adds: "I would say that you're not going to lose your job to artificial intelligence, but there might be the possibility you're going to lose your job to someone who knows how to use artificial intelligence if you don't."
