"Engineering Trends: Future Work or Fleeting Fashion?"

2023-08-29

With the rise of generative AI, there has been an increasing number of courses and certifications to help people enter the "hot job" of prompt engineering.

Using natural language, such as English, to "prompt" AI models like ChatGPT and Midjourney to generate useful content seems to be highly valued by many employers. But can you really earn a six-figure salary by taking a short-term course?

The Hype of Prompt Engineering

An article published by The Washington Post in February of this year largely planted the idea that prompt engineers are "AI whisperers" who "program with prose." The article threw out some high salary figures and quoted a recruitment advertisement from Silicon Valley company Anthropic, seeking individuals with a "creative hacker spirit" and a love for puzzles.

Similar articles in Time, Forbes, and Business Insider further fueled this frenzy.

To complete the transition from geek to trendy, several influential figures joined in, portraying prompt engineering as a gold rush where anyone willing to learn and master a few skills can find a job.

Are There Really That Many Jobs?

That Anthropic advertisement is still out there. Six months later, it looks more like a corporate gimmick than a talent search.

As predicted by many commentators, prompt engineering has not emerged as a standalone profession. At the time of writing this article, there were no job advertisements for "prompt engineers" on major Australian job websites. Only four job ads mentioned prompt engineering in their job descriptions.

The situation seems slightly better in the United States. However, even there, this new profession is largely categorized under other roles such as machine learning engineer or AI specialist.

There is almost no reliable statistical data on the growth (or lack thereof) of prompt engineering. Most of the data is hearsay. Consulting firms like Deloitte include it as part of the AI business development and promote it as the "dawn of a new era," further obscuring the reality.

What Is the Reality?

Many of the misconceptions about the usefulness of prompt engineering stem from not recognizing the existence of two different types of value creators: domain experts and technical experts.

Domain Experts

The truth behind the claim that "anyone can do it" lies in the fact that domain experts in a particular field are often the best prompters for accomplishing specific tasks. They know what questions to ask and can find value in the answers.

For example, in the field of branding and marketing, generative AI shines in general or "G-type" creative tasks, such as creating a Pepsi logo in the style of Picasso. When advertising experts start prompting, they quickly come up with ways to do things that even the most skilled AI masters cannot. This is because technical experts often lack knowledge of copyright or marketing.

Technical Experts

On the other hand, technical masters who deal with complex AI models can also serve as prompt engineers and add value to them. They understand the secrets of AI models.

For example, they can use this knowledge to improve the results of using AI to extract data from internal company documents. However, they often have little knowledge of domains outside of AI.

Both domain expert and technical expert prompt engineers are valuable, but they have different skills and goals. If an organization is using generative AI on a large scale, it may need both.

Why Is Prompting Difficult?

Generative AI provides output for people. The usefulness or uselessness of ad copy, images, or poetry can only be judged after they succeed or fail in the real world. In many real-world scenarios, domain experts are the only ones who can judge the usefulness of AI output.

However, these evaluations are ultimately subjective. We know that 2 + 2 = 4. So, testing to prevent AI from mistakenly thinking the answer is 5 is simple. But how long does it take to determine if an AI-designed advertising campaign is more effective than one designed by a human (even with a domain expert)?

Is Prompt Engineering Worth Learning?

Aside from playing with some tricks and techniques, formal learning on how to write prompts seems somewhat meaningless for most people. Firstly, AI models are constantly updated and replaced. What works as an effective prompt now may only be effective in the short term.

For those hoping to get rich quick with prompt engineering, it is best to focus on combining AI with their own domain expertise. For example, if you are a pharmacist, you might try using generative AI to carefully check warning labels on prescriptions.

In the process, you can hone your explanatory writing skills, gain basic generative AI skills (which employers may appreciate), and potentially create killer applications for the right audience.

In the end, boasting about knowing how to prompt AI will not appear on a resume. It is akin to boasting about knowing how to use a search engine.

RECENT AI NEWS

RECENT AI TOOLS