Ask HN: Prompt Engineering: A Skill or Role? Whats the Future?

I am seeking insights into the future of Prompt Engineering across different segments:

- Company Sizes: Small, Mid-Size, Large - Roles: Tech, Product, Non-Tech

How much time do you foresee developers or companies dedicating to prompt engineering?

I understand this question is quite open-ended, but I'm eager to gather perspectives.

I suspect that the more intricate/technical the skill is, plus the bigger the ROI of that extra technical know-how then the more likely a role will exist for this. So for example, if any person can get the same results as a dedicated prompt eng, then the less likely a prompt engineer role will exist. Also, if the prompt eng can get very different (and higher quality results) than any person, but the results are not that profitable, then less likely a prompt engineer roles will exist.

Bonus points for the likelihood of a dedicated prompt engineer role if the engineer can apply their wizardry across many aspects of the company. This kind of reminds me of the OSS (operational systems support) team I used to be part of, and still miss. In this team we applied mostly custom software solutions to any team that needed it. We worked hard to find customers within the company to earn our keep, else we would get laid off at the next convenient chance. I imagine prompt engineers, if given the chance, should work hard to find value and not just wait for the work to come to them.

I think that agents will be fairly ubiquitous within a few years. So instructing AI will be a critical skill.

There may also be a very large segment of the population that are small business entrepreneurs with only AI and/or robotic "employees".

There will be a very popular category of AI/robotic "co-founder" agent that does the "prompt engineering" for you to set up and reconfigure your business by selecting the team of bots and specifying instructions for them.

We will also start to see more and more anticipatory companion agents that somehow monitor the stream of input that you consume as well as your output and just predict and produce much of what you want before you request it.

I personally don't see a big future in traditional jobs.

It is already starting. But most of those predictions may be widely applicable in 2-10 years.

It seems like all of these things will be feasible to some degree within a couple of years. It's hard to predict how long it will take for them to be robust and widely deployed.

Scrum master is a defined role though. What makes you say it's bloated?