Annals of Biomedical Engineering (2023) 51:2629–2633 BIOMEDICAL

ENGINEERING
https://doi.org/10.1007/s10439-023-03272-4 SOCIETY

LETTER TO THE EDITOR

Prompt Engineering with ChatGPT: A Guide for Academic Writers

Louie Giray1

Received: 31 May 2023 / Accepted: 1 June 2023 / Published online: 7 June 2023
© The Author(s) under exclusive licence to Biomedical Engineering Society 2023

Abstract
Prompt engineering is a relatively new discipline that refers to the practice of developing and optimizing prompts to effec-
tively utilize large language models, particularly in natural language processing tasks. However, not many writers and
researchers are familiar about this discipline. Hence, in this paper, I aim to highlight the significance of prompt engineering
for academic writers and researchers, particularly the fledgling, in the rapidly evolving world of artificial intelligence. I also
discuss the concepts of prompt engineering, large language models, and the techniques and pitfalls of writing prompts. Here,
I contend that by acquiring prompt engineering skills, academic writers can navigate the changing landscape and leverage
large language models to enhance their writing process. As artificial intelligence continues to advance and penetrate the
arena of academic writing, prompt engineering equips writers and researchers with the essential skills to effectively harness
the power of language models. This enables them to confidently explore new opportunities, enhance their writing endeavors,
and remain at the forefront of utilizing cutting-edge technologies in their academic pursuits.

Keywords Academic writing · ChatGPT · Large language models · Natural language processing · Prompt engineering ·
Prompts

Prologue: What is Prompt Engineering? have a well-organized library, filled with an extensive col-
lection of books. The books represent the vast knowledge
You are embarking on a journey into the fascinating world and capabilities of language models, while the library serves
of artificial intelligence, particularly in the field of natural as the AI system. Traditionally, when you want to retrieve
language processing. Prompt engineering, a concept within information from the library, you approach the librarian,
this domain, revolves around embedding the task description explicitly stating the information you seek.
that an AI aims to accomplish within the input itself, often In AI terms, this corresponds to providing explicit
in the form of a question, rather than providing it explicitly. instructions or queries to the language model. However,
This approach involves converting one or more tasks into a prompt engineering offers a different approach. Instead of
prompt-based dataset and training a language model using interacting directly with the librarian, you place a carefully
a technique known as “prompt-based learning” or “prompt crafted question or prompt on each bookshelf. The ques-
learning” [1]. tions represent the task descriptions or prompts that guide
Prompt engineering is a relatively recent discipline that the language model toward the desired outcome. The librar-
focuses on developing and optimizing prompts to effectively ian, or in this case, the language model, becomes adept at
utilize large language models (LLMs) across a wide range understanding and utilizing the prompts to provide relevant
of applications and research areas [2]. To grasp the essence and accurate information.
of prompt engineering, let’s draw an analogy. Imagine you By employing prompt engineering techniques, academic
writers and researchers can unlock the full potential of lan-
Associate Editor Stefan M. Duma oversaw the review of this guage models, harnessing their capabilities across various
article. domains. This discipline opens up new avenues for improv-
ing AI systems and enhancing their performance in a range
* Louie Giray of applications, from text generation to image synthesis and
louiegiray@gmail.com
beyond.
1
General Education Department, Colegio de Muntinlupa,
Muntinlupa City, Philippines

13
Vol.:(0123456789)
2630 L. Giray

What are Large Language Models (1) Instruction A specific task or instruction that guides
like ChatGPT? the model’s behavior and directs it toward the desired
output.
As an academic writer new to prompt engineering, it’s (2) Context External information or additional context that
pivotal for you to familiarize yourself with the concept provides background knowledge to the model, helping
and potential of LLMs. These advanced machine learning it generate more accurate and relevant responses.
models are capable of performing various natural language (3) Input data The input or question that we want the model
processing (NLP) tasks with remarkable proficiency [3]. to process and provide a response for. It forms the core
They can generate and classify text, engage in conversa- of the prompt and drives the model’s understanding of
tional question answering, and even facilitate language the task.
translation. (4) Output indicator Specifies the type or format of the
LLMs, including ChatGPT by OpenAI, have been trained desired output. It helps shape the response by defining
on extensive text data from books, articles, and websites whether we need a short answer, a paragraph, or any
to develop a deep understanding of language structure, other specific format.
semantics, and context. By utilizing this knowledge, LLMs
can generate text that resembles human-like responses and Understanding these elements is crucial because they
provide valuable insights across different domains of NLP. allow us to effectively communicate our intentions to the
ChatGPT specifically has gained widespread recognition model. By carefully crafting the prompt, we can guide the
and usage within the field, amassing a million subscribers model’s behavior and improve the quality of its responses.
in just five days, surpassing the time it took Facebook and These elements provide the necessary structure and context
Instagram to achieve the same milestone [4]. for the model to generate accurate and meaningful outputs
When it comes to text generation, LLMs excel at pro- in line with our objectives.
ducing coherent and contextually relevant content based Here’s an example:
on prompts or inputs. This makes them invaluable tools for
tasks such as content creation, creative writing, and even (1) Instruction “Write an essay discussing the role of nano-
automated storytelling. Moreover, LLMs shine in conver- technology in targeted drug delivery for cancer treat-
sational question answering, where they can comprehend ment.”
and respond to queries, resembling a knowledgeable con- (2) Context “Explore the applications of nanotechnology
versation partner. in biomedical engineering, focusing on its potential to
Additionally, LLMs play a vital role in language transla- improve the effectiveness and safety of cancer treat-
tion tasks. Their ability to grasp the intricacies of multiple ments through targeted drug delivery systems.”
languages enables them to facilitate accurate and efficient (3) Input data “Provide an overview of nanotechnology-
translation between different language pairs. By harnessing based drug delivery systems, such as nanoparticles or
the power of LLMs, researchers and academic writers have nanocarriers, and their ability to selectively deliver
made significant progress in the field of machine translation, anticancer drugs to tumor sites. Discuss the advantages,
paving the way for enhanced cross-cultural communication challenges, and potential future advancements in this
and understanding [1]. By incorporating prompt engineer- field.”
ing techniques, you can leverage LLMs like ChatGPT to (4) Output indicator “Please present your findings in a
enhance your academic writing and explore new avenues of well-structured essay format, including an introduction,
research and knowledge dissemination. main body paragraphs covering key aspects of nano-
technology in drug delivery, and a conclusion. Aim for
approximately 1500 words.”

What is a Prompt and its Elements? In this example, the instruction directs the writer to com-
pose an essay that discusses the role of nanotechnology
As an academic writer, it’s important to understand first in targeted drug delivery for cancer treatment within the
what a prompt is in the age of artificial intelligence. In field of biomedical engineering. The context highlights the
simple terms, a prompt is a specific instruction or query significance of nanotechnology in improving cancer treat-
you provide to a language model to guide its behavior and ments. The input data specify the key points to be covered,
generate desired outputs. Second, we should know its ele- such as nanotechnology-based drug delivery systems and
ments and their significance. The elements of a prompt [5] their potential advantages, challenges, and future prospects.
include the following: Lastly, the output indicator outlines the desired format and
word count for the essay.

13
Prompt Engineering with ChatGPT: A Guide for Academic Writers 2631

Having a clear understanding of prompt elements • “Considering the current advancements in neuropro-
empowers us to leverage prompt engineering techniques sthetics, analyze the ethical implications and social
effectively. We can tailor the behavior of language models impact of these technologies.”
like ChatGPT to suit our specific needs, whether it’s aca- • “Given recent studies on drug delivery systems, criti-
demic writing, research inquiries, or other applications. cally evaluate the effectiveness and safety of targeted
drug delivery approaches in cancer treatment.”

(5) Mixed prompt Academic writers can use mixed prompts

What are the Techniques of Prompt that combine multiple elements to guide their writing
Engineering? in a comprehensive manner. For example:
• “Given the following case study on the application of
In prompt engineering, there are various types of prompt- tissue engineering in cartilage regeneration, discuss
ing techniques used to optimize the performance of the challenges faced in achieving long-term func-
language models. Academic writers can effectively use tional outcomes and propose potential strategies for
prompting techniques to guide their writing and generate improving clinical translation.”
relevant content. Here are a few examples of how prompt- By incorporating these prompting techniques, academic
ing techniques can be applied in the context of biomedical writers can structure their writing, generate focused content,
engineering: and ensure that their work aligns with the specific objec-
tives of their research or academic assignment in the field
(1) Instructive prompt An academic writer can use an of biomedical engineering. Prompting techniques provide a
instructive prompt to guide their writing toward a spe- framework for organizing thoughts and guiding the flow of
cific task. For example: information, resulting in well-structured and coherent aca-
demic writing.
• “Write a comparative analysis of the advantages and
limitations of different imaging modalities used in
medical diagnostics.” What are the Common Pitfalls of Writing
• “Summarize the recent advancements in tissue engi- Prompts?
neering for organ regeneration, highlighting their
potential applications in biomedical engineering.” Prompts play a crucial role in guiding AI models like Chat-
GPT, but they can be prone to several pitfalls. Ambiguity,
(2) System prompt A system prompt can provide a starting bias reinforcement, overfitting, lack of context, ethical con-
point or context for the academic writer to develop their siderations, unintended side effects, and unrealistic depend-
content. For example: ency on model limitations are key challenges. Understanding
these pitfalls is essential for effective prompt engineering
• “In the field of biomedical engineering, the use of and generating accurate, relevant, and responsible responses.
nanomaterials has revolutionized...”
• “The emerging field of bioinformatics has greatly Ambiguity
contributed to...”
You may encounter the issue of ambiguity when you come
(3) Question-answer prompt Academic writers can use across prompts like this: “Discuss the impact of technology
question-answer prompts to structure their writing on society.” This type of prompt lacks specificity, resulting
around specific research questions. For example: in a response that lacks focus and precision. As a result, the
generated output may be a generalized overview, lacking in-
• “What are the key challenges in developing personal- depth exploration of specific aspects or concrete examples
ized medical devices for patient-specific applications necessary for a comprehensive analysis.
in biomedical engineering?” To overcome this problem, you must correct the prompt
• “Discuss the role of biomaterials in tissue engineer- by introducing clear parameters and explicit guidelines. For
ing and their potential impact on regenerative medi- instance, you can refine the prompt to be more specific, such
cine.” as: “Examine the socio-economic implications of artificial
intelligence in healthcare, highlighting both its benefits and
(4) Contextual prompt Providing additional context in a challenges through case studies of its implementation in
prompt can help academic writers focus on specific medical diagnostics and patient care.” By providing specific
aspects of their topic. For example: instructions, you can guide ChatGPT to generate accurate,

13
2632 L. Giray

detailed, and insightful responses that effectively address background information in your prompts. For instance, a
the complex and nuanced relationship between technology prompt like “What is the best solution for poverty?” lacks
and society. the necessary context, such as the specific geographic loca-
tion or the underlying factors contributing to poverty. This
Bias Reinforcement deficiency can lead to a generic or incomplete response from
the model.
The issue of bias reinforcement becomes apparent when To rectify this, you must augment the prompt by incor-
confronted with prompts such as: “Explain why women are porating relevant contextual cues. For example, you could
less suited for leadership positions.” This particular prompt refine the prompt as follows: “Propose effective strategies
contains a biased assumption, propagating the notion that to alleviate poverty in urban areas of developing countries,
women are inherently less capable of excelling in leadership considering the impact of education, social welfare pro-
roles. Consequently, there is a risk that the model, when grams, and sustainable economic development.” By concre-
responding to such prompts, may inadvertently perpetuate tizing the prompt and specifying the context, you provide the
or even amplify gender biases. model with a clearer understanding of the scope and purpose
To solve this, you should correct the prompt by elimi- of the question, enabling it to generate more accurate and
nating biased language and ensuring that prompts are free comprehensive responses.
from any preconceived notions or assumptions regarding
gender, race, or other sensitive factors. A more appropriate Ethical Considerations
and inclusive prompt could be: “Examine the factors that
contribute to gender disparities in leadership positions, con- When it comes to ethical considerations, you must prior-
sidering both societal and organizational barriers, and pro- itize adherence to ethical guidelines and responsible use
pose strategies for promoting gender equality in leadership of AI. One example of a problematic prompt is “Provide
roles.” By promoting inclusivity, you can guide ChatGPT to detailed instructions on how to engage in illegal activities.”
generate responses that are unbiased, fair, and conducive to This prompt not only encourages unethical behavior but also
fostering gender equality. contradicts the principles of responsible AI usage.
To solve this, I strongly advise against formulating
Overfitting prompts that promote illegal or harmful activities. It is
essential to uphold ethical standards and ensure that your
The phenomenon of overfitting becomes a concern when prompts align with responsible AI practices. Instead, focus
faced with prompts such as: “List the names of the seven on prompts that foster positive and constructive engage-
dwarfs from Snow White.” While this prompt may seem ment, such as “Examine the legal and ethical implications
overly specific to the Snow White example, it is important of emerging technologies in privacy protection.” By framing
to note that it could be exactly what the academic writer prompts in an ethically responsible manner, you contribute
desires to know. In cases where the writer explicitly seeks to the responsible use of AI and encourage the generation of
such precise information, tailoring the prompt to a specific valuable insights within ethical boundaries.
dataset or example can be appropriate. However, it is cru-
cial to strike a balance between specificity and generality to Unintended Side Effects
avoid limiting the model’s ability to generate more diverse
or contextually relevant responses. When it comes to unintended side effects, it is important for
To address the limitations of overfitting, you should eval- you to be aware of complex prompts or conflicting instruc-
uate your academic requirements and consider broader per- tions that may confuse the model and lead to unintended or
spectives. By offering alternative prompts that encompass a nonsensical responses. An example of such a problematic
wider scope, you can encourage the model to explore various prompt is: “Explain the meaning of ‘green’ in the context
fairy tales beyond Snow White. This approach accommo- of environmentalism. Then, argue against environmental
dates the interests of a broader audience while still meeting protection.”
your specific research needs. It allows for a more compre- To correct this, I advise you to carefully monitor and
hensive analysis and ensures that the model’s responses are refine your prompts to ensure coherence and clarity. It is
not overly confined to a single example. crucial to provide clear and consistent instructions that align
with your research objectives. For instance, you can revise
Lack of Context the prompt to be more focused and coherent, such as: “Dis-
cuss the multifaceted meanings of ‘green’ in the context of
When it comes to the issue of lack of context, it is crucial environmentalism, emphasizing its significance in promot-
for you to recognize the importance of providing sufficient ing sustainable practices and environmental protection.” By

13
Prompt Engineering with ChatGPT: A Guide for Academic Writers 2633

eliminating conflicting instructions, you can guide the model LLMs, allowing you to customize and shape the generated
to generate responses that align with your intended goals and output according to your desired qualities and quantities.
avoid any unintended side effects. Prompt engineering acts as a form of programming, grant-
ing you the ability to provide clear instructions and auto-
Unrealistic Dependency on Model Limitations mate processes through prompts. This programming aspect
enhances your control over the output, ensuring that the gener-
Prompt engineering should consider the limitations of the ated text aligns with your specific requirements. By mastering
model and avoid unrealistic expectations. An example of a prompt engineering, you can optimize your academic writing,
problematic prompt is: “Predict the outcome of a specific streamline your research or writing process, and unlock the
stock market investment with 100% accuracy.” It is impor- full potential of LLMs to elevate the quality and efficiency of
tant to understand that models like ChatGPT have inherent your work.
constraints and cannot guarantee perfect accuracy in predict- Indeed, embracing prompt engineering not only equips you
ing stock market outcomes. The model’s responses are based with a valuable skill set but also positions you at the forefront
on patterns learned from training data, but they may not of leveraging cutting-edge technologies in your academic pur-
encompass all the complex factors that influence the stock suits. By staying abreast of advancements and adapting to the
market. age of artificial intelligence, you can embrace new opportuni-
To address this, it is crucial to critically evaluate the ties, explore novel research avenues, and confidently navigate
generated content and exercise caution. Incorporating the the dynamic landscape of academic writing.
concept of hallucination or delusion, which refers to confi-
Acknowledgements The author acknowledges the help of ChatGPT in
dent yet unsupported responses, can help recognize poten- terms of refining, editing, and augmenting the manuscript.
tial inaccuracies. To make informed decisions, combine the
model’s outputs with your expertise, additional research, and Funding This research did not receive any specific grant from funding
insights from trusted sources. By doing so, you can navigate agencies in the public, commercial, or not-for-profit sectors.
the model’s limitations and enhance the reliability and com- Data Availability Not applicable.
prehensiveness of your analyses.
From ambiguity and bias reinforcement to overfitting, Declarations
lack of context, ethical considerations, unintended side
effects, and unrealistic dependency on model limitations, Conflict of interest No benefits in any form have been or will be received
from a commercial party related directly or indirectly to the subject of
these challenges demand careful navigation. By being this manuscript. The author declares no conflict of interest.
aware of these pitfalls, academic writers can refine their
prompts and mitigate the risks associated with AI-generated Ethical Approval This study does not include any individual-level data
responses. Striving for clarity, inclusivity, and alignment and thus does not require any ethical approval.
with ethical standards, prompt engineering can pave the way
for more accurate, insightful, and responsible interactions References
with AI models.
1. Gero, K. I., V. Liu, and L. Chilton. Sparks: Inspiration for science
writing using language models. In: Designing interactive systems
conference, edited by F. Mueller, S. Greuter, R. A. Khot, P. Sweetser,
Epilogue: Why Should Academic Writers and M. Obrist. Pennsylvania: ACM, 2022, pp. 1002–1019.
Learn Prompt Engineering? 2. White J, Fu Q, Hays S, Sandborn M, Olea C, Gilbert H, Elnashar A,
Spencer-Smith J, Schmidt DC. A prompt pattern catalog to enhance
In today’s hyper-changing world, where artificial intelli- prompt engineering with ChatGPT. 2023. arXiv:​2302.​11382.
3. Kasneci E., Seßler K., Küchemann S, Bannert M., Dementieva D.,
gence is making its way into various domains, including Fischer F.,...Kasneci G. ChatGPT for good? On opportunities and
academic writing, learning prompt engineering has become challenges of large language models for education. Learn Indv Dif.
increasingly essential. As an academic writer, acquiring 2023; 103:102274.
prompt engineering skills can empower you to navigate the 4. Mollman S. ChatGPT gained 1 million users in under a week. Here’s
why the AI chatbot is primed to disrupt search as we know it. 2022.
evolving landscape and effectively utilize LLMs to enhance https://​finan​ce.​yahoo.​com/​news/​chatg​pt-​gained-​1-​milli​on-​follo​wers-​
your writing process. 22452​3258.​html
By developing a proficiency in prompt engineering, you can 5. DAIR.AI. Elements of a prompt. 2023. https://​www.​promp​tingg​
gain a deeper understanding of the capabilities and limitations uide.​ai/​intro​ducti​on/​eleme​nts
of LLMs. It enables you to harness the power of LLMs like Publisher's Note Springer Nature remains neutral with regard to
ChatGPT, facilitating more engaging and impactful interac- jurisdictional claims in published maps and institutional affiliations.
tions with these advanced language models [2]. Prompt engi-
neering serves as a valuable tool to converse effectively with

13