The Emergence of Prompt Engineering: Bridging Linguistics and AI

Unlocking the Power of Language Models through Strategic Prompts

Prompt engineering has emerged as a new role in the rapidly evolving field of AI, and surprisingly, you don't necessarily need a computer science background to excel in this position. Take the case of Teodora Danilovic, who became an AI-prompt engineer after completing a degree in philosophy. She leveraged her knowledge of logic, linguistics, and critical thinking to excel in this field, and acquired technical skills on the job.

Teodora joined AutogenAI, a company that has developed a language model called "Genny" for bid and tender writing. As a prompt engineer, her primary responsibility is to identify the best inputs to ensure users receive optimal responses from the software. These inputs could involve tasks such as expanding text, summarizing content to specific word counts, or incorporating crucial evidence in bids. AutogenAI's software provides various functions like rephrase, summarize, and tone of voice, enabling users to adapt their own text accordingly.

Teodora's role involves ensuring that the coding and prompts behind these functions yield accurate and consistent results. Additionally, she collaborates with engineers to adapt the coding and prompts as the AI system evolves. While generative AI may seem like a purely technical field, Teodora discovered that it incorporates a significant linguistic element, making it a perfect fit for her expertise in language usage and nuances.

Teodora's background in analytic philosophy, including mathematical logic, formal logic, philosophy of science, and linguistics, equipped her with a deep understanding of language. She recognizes that language is not merely a means of exchanging ideas, but a dynamic force with tangible impacts on the world. Her experience working at a cryptocurrency company and studying the ethics of AI during her degree further enhanced her understanding of the field.

Teodora emphasizes the importance of crafting unambiguous, direct, and relevant prompts for language models. Users often encounter problems when assuming that models can understand intuitive social, contextual, and intentional cues. To avoid such issues, prompts need to provide explicit instructions rather than relying on intuition. For example, asking the language model to "improve the paragraph above by removing all grammatical errors" is more effective than a vague command like "make this paragraph better."

Ensuring consistency, optimization, and diversity in the outputs of AutogenAI's software features is an ongoing process of improvement. With the release of ChatGPT-4, prompt engineers like Teodora had to retest existing prompts to ensure compatibility with the new model. While ChatGPT-4 offers advanced capabilities, such as following detailed instructions, it tends to provide verbose responses, which may not be suitable for bids, tenders, or proposals that require concise summaries.

Teodora highlights that being a prompt engineer requires not only an understanding of language but also a technical comprehension of the language model and its architecture. Although she is not a developer, she has gained basic programming skills on the job, enabling her to suggest and make edits to the code. Furthermore, she has developed knowledge of natural language processing and machine learning within the AI domain, along with an understanding of the research and development process involved in building language models.

While prompt engineers in Silicon Valley reportedly earn substantial salaries, Teodora acknowledges that her compensation is respectable, although not at the level of those high figures. Remuneration typically increases with experience and expertise, as well as the range of skills and disciplines one possesses. Instead of fearing job displacement by AI, Teodora believes that AI tools like AutogenAI empower clients to focus on more engaging tasks by automating mundane activities.

Prompt engineering represents just one of the novel roles that have emerged in the rapidly evolving AI landscape. It's a role that individuals with suitable skills, such as critical thinking, analytic questioning, and linguistic expertise, may already possess, making it a potentially rewarding career option.