Will the engineer Prompt be the new "sexiest job of the 21st century"?

In 2012, Thomas H. Davenport and DJ Patil called data science "the sexiest job of the 21st century" (Harvard Business Review). That may have been true at the time. And even today, deep expertise in building AI-based systems remains a highly sought-after (and in short supply) skill. Since that statement was made over 10 years ago, the landscape and tools of data science have changed dramatically. Developers with a basic understanding of data science can create models that are "good enough"—good enough to get started or even to be put into production. While data scientists are still needed for complex AI projects, this evolution makes AI more accessible to a wider range of people. A similar trend may also emerge in the coming years in a different area of ​​the AI ​​space: instructing generative AI systems to create a desired output—text, image, audio/video, and so on. – with minimal effort and high precision. This task is called “prompt engineering”.

Current generative AI systems work by asking the user to enter what's called a "prompt." This is a set of commands that instruct the AI ​​system to generate a result. For example: "Imagine you are a technical writer. Describe what this code does in English," or "Write a blog post about the three best cities to host an AI conference. Keep it under 2000 words. Use a professional tone." If it's written in plain English, what is the job then?

Using generative AI applications is a lot like experimentation. It may take a long time for the system to generate what you want. For example, iterating over a text prompt and making small changes before the result matches your intent. To get results quickly, it's essential to understand what you want to achieve as an end result and to know how to articulate it in plain language. The iterations of a prompt can be quite small, for example changing a word, but they will probably change the output generated. You just don't know exactly how until you submit your application. But generative AI systems will evolve and become easier to use.

In a recent interview, Sam Altman (CEO of OpenAI, the company behind ChatGPT) stated that prompt engineering will no longer be necessary within the next five years (YouTube). If this is the case, how can we use AI?

I currently see three scenarios for the evolution of the use of generative AI:

Personalization: AI systems will know enough about their users and their preferences and need less context.

Focus: Generative AI will move from product status to mere product functionality and require fewer explicit instructions.

User Experience: The way we interact with AI will be less like coding and, at a minimum, like using our current user interfaces.

What can AI leaders do now?
The renewed interest in AI is largely due to its mainstream appeal and ease of use, in addition to its unprecedented capabilities. Although some training and experimentation is required, users do not need to become experts in statistics and mathematical optimization to use it. This makes generative AI much more accessible compared to traditional data science where models are built from the ground up. It also creates opportunities for companies to benefit from AI-based innovation faster and with a much shorter learning curve. AI and technology leaders can further support this move and:

Encourage their teams to experiment with available generative AI systems,

inform them on how they can try them out safely (challenges/pitfalls, information security),

provide hands-on opportunities (hackathons) and compare grades and results,

and embrace this evolution of AI. For now, prompt engineering could succeed data science as the "sexiest" job of the decade.

How do you refine your "Prompt Engineering" skills?

To delve deeper: this question ties in with our article on the job of Prompt Designer and our 100 examples of roles that AI can play.

source: https://intelligencebriefing.substack.com/p/will-prompt-engineer-be-the-sexiest