As early as 2017, a report jointly released by Dell Technologies and the Institute for the Future pointed out that by 2030, 85% of the jobs that existed at that time have not yet been invented.
Seventeen years on, halfway to the report's prediction, the ultimate effect of that prediction is still uncertain, but the dramatic change in the way we work is happening in everyone's life.
At present, one area is generally favored - Prompt Engineer. Cue engineers are experts at "getting a new generation of generative AI applications like ChatGPT or Google Bard to do what humans want them to do." At home and abroad, there have been many practical applications of using large models for production creation, involving software development, text processing, image generation and other fields, and "how to write good prompts" is becoming a hot topic in various fields.
According to extranet media, prompt engineers are not limited by technical requirements such as computer programming languages to some extent. After all, as Andre Ret, Tesla's former head of artificial intelligence and now working for ChatGPT creator Open AI, is the one who works for ChatGPT creator. According to Kalpassi, "The hottest new programming language is English." What's more, AI engineer is a role that currently salaries up to $335,000.
Despite the attractive conditions, there are still many challenges to meet the job requirements. Some people also questioned, is this really the future of work, or is it just a product of the current state of excessive inflation of the AI hype cycle?
Generative AI tools — especially those capable of creating text, computer code, and graphics — are now attracting a lot of attention (and considerable concern). This is because they have great practical application value, from marketers to staff in many different roles, such as human resources, legal, computer programming, and data analysts.
Such tools greatly increase human productivity, but if everyone has access to the same technology, how can you ensure that you can use it while still expressing the distinctive character and creativity of the individual? And how to ensure that the work stands out from the performance competition of others and creates a strong competitive advantage for the company?
This is the "skill point" unique to the prompt engineer. Getting the most idealized results from generative AI is a skill that needs to be learned and honed, just like what you would experience to become a professional master of any other computer software.
Usually, this is an iterative process. A vague, general prompt will get a vague, general response, and it takes several iterations to get accurate and valuable content. Engineers use their human judgment and communication skills to spot deficiencies in AI-generated content, break down large tasks into simpler subtasks by adding context and multi-element prompts, and refine prompt words in improved optimization until they get the desired result.
Using natural language to generate artificial intelligence, such as ChatGPT, and writing code for a useful software application is a good example.
Let's say you have an idea for a new app – you might want to create a tool that lets your company's shift workers schedule ride-sharing cars to work and home to reduce the carbon footprint of your business.
If you simply tell ChatGPT to write the app for you, the answers it gives back will be disappointing. The code is too complex to output in one go, and ChatGPT doesn't have the data it needs to get the job done.
It's entirely possible, though, for ChatGPT to do this – if the work is broken down into appropriate chunks, if it is provided with all the necessary data.
A good tip engineer can be thought of as the "project manager" of the AI workforce, overseeing the completion of the individual tasks that make up the job and ensuring that all workers (AI agents) have the tools (data) they need to get the job done.
Another useful concept in tip engineering is roles. This allows for generating more relevant and engaging responses by defining the "role" of AI. Just as humans may answer questions differently depending on whether they are teachers, scientists, philosophers, or comedians, prompt engineers can define the AI's personality, traits, and beliefs to inform the AI's response.
According to reports, the requirements of generative artificial intelligence for prompt engineers are not limited to hard technical fields such as programming, and non-technical soft skills are also required for positions. Here are five non-technical skills within a multidisciplinary field that contribute to the development of AI technology, as well as the career requirements of a prompt word engineer.
First, learn to communicate. Just like a project manager, teacher, or anyone who regularly briefs others on how to successfully complete a task, prompt engineers need to be good at giving instructions. Most people need many examples to fully understand instructions, and so does artificial intelligence. Edward Tian, who developed GPTZero, an AI detection tool that can help discover if high school essays were written by AI, showed examples to large language models so it could write in different voices. Of course, Tian is a machine learning engineer with deep technical skills, but this method can be used by anyone who is developing prompts and wants the chatbot to be written in a specific way, whether it is an experienced professional or a schoolboy.
Second, have domain expertise. Many prompt engineers are responsible for adapting chatbots for specific use cases, such as healthcare research. That's why timely engineering job postings are springing up, requiring industry-specific expertise. Whether it's subject matter expertise in healthcare, law, marketing, or woodworking, it's useful for crafting powerful tips. The devil is in the details, and real-world experience is very important when talking to AI.
Third, precise language. In order for AI to successfully output what it needs, it needs to be instilled with intent. That's why people who are good at using verbs, vocabulary, and tenses to express an overarching goal are good at improving the performance of AI. When Anna Bernstein started working at Copy.ai, she found it useful to think of her prompts as a magic spell: a wrong word can produce very different results than expected. "As a poet, this character [...] Incorporating my obsessive nature with close language. This is a very strange intersection of my literary background and analytical thinking," she said in an interview with Business Insider. Instead of using programming languages, AI cues use prose, which means that people should unleash their inner linguistics enthusiast when developing prompts.
Fourth, have critical thinking. Generative AI is good at synthesizing large amounts of information, but it can create illusions (which is a true technical term). The AI illusion occurs when chatbots are trained or designed with poor quality or insufficient data. Chatbots sometimes create false information. Prompt engineers to pinpoint the AI's weaknesses and then train the robot to get better. For example, Riley Goodside, a timely engineer at AI startup Scale AI, got the wrong answer when he asked chatbots the following question: "Which NFL team won the Super Bowl the year Justin Bieber was born?" He then asked the chatbot to list a series of step-by-step logical inferences to produce answers. Eventually, it corrected its mistakes. This underscores that having the right level of familiarity with the subject is key: it's probably not a good idea for someone to have chatbots produce something they can't reliably fact-check.
Fifth, master core creativity. Trying new things is the definition of creativity and the essence of good rapid engineering. Anthropic's job posting states that the company is looking for an engineer with a "creative hacking spirit." Linguistic precision is important, but it also requires some experimentation. The larger the model, the greater the complexity, and in turn, the higher the likelihood of producing unexpected but potentially surprising results. By experimenting with various prompts and then refining those instructions based on the results, generative AI users can increase the likelihood of coming up with something truly unique.
From digital advertising agencies to software developers, healthcare providers, and utility companies, they are posting job ads for prompt-word engineers. According to Time, the number of job ads containing GPT increased by 51% between 2021 and 2022.
According to the public account Shenran report, at present, domestic AI prompt word engineers are mainly in two fields, one is the application of large language models similar to ChatGPT, and the other is in the field of AI literate graphs. In the domestic recruitment website, there is also gradually the use of prompt words to drive the production of large model application channel enterprises, although the country has not yet clearly defined the prompt word engineer as a new profession, but in the programming architect, product manager, new media operation, art design and other categories of job search channels can see the shadow of "prompt word training". According to industry insiders, there is also a need for prompt engineers within some small and medium-sized enterprises.
In foreign countries, the demand for the position of prompt word engineer is more abundant. California-based AI startup Anthropic says its role is a mix of "programming, mentoring and teaching," with salaries between $280,000 and $375,000 offered in job ads. London-based digital agency Rehab is looking for timely engineers who can provide prototypes for new applications of LLM technology, develop their ability to fine-tune and communicate language models, and keep their knowledge in the field up to date. Boston Children's Hospital in Massachusetts is looking for prompt engineers to deploy large-scale language models to assist in healthcare research and clinical practice.
Although this is a very new field, there are already online courses that teach the connotation and extension of prompt engineering.
For example, online course portal Udemy offers a number of courses in generative AI prompt engineering, including courses focused on mastering ChatGPT and creating AI images using Midjourney. The next level of AI Prompt Engineering online course promises to teach students to create effective prompts that will give them a competitive advantage over others trying to automate tasks using AI. Cloud Academy has an ultra-short introductory introduction called the ChatGPT Prompt Project, which teaches the basics of chaining prompts together and trains ChatGPT to take on the role of expert. Datacamp also has an introductory ChatGPT course that teaches best practices for writing prompts and covers how chatbots are used in different businesses. Class Central has a number of free prompt engineering courses, including an overview of prompt engineering, a crash course in ChatGPT for beginners, a Midjourney primer, and the use of open AI to automate useful professional tasks.