Prompts // 6 (Conscientious Commands)

I can’t seem to leave this alone, and this sixth post in the series on understanding and learning prompt engineering will focus on the sometimes overlooked consideration of ethics – in using generative AI and Large Language Models (LLMs) as source material, using it to inform decisions, if (or when) students use it for their learning or assessment, etc.

There are many aspects to the ethical considerations in prompt engineering – beneath the surface of the interaction with AI lies a complex weave of ethical data. As leaders, educators and technologists, it is within our reach to make sure we understand the importance of our instructions as we create the prompts that guide AI.

In this blog post, I will look to embark on a reflective journey through the ethical implications of prompt engineering, emphasising the profound responsibility we carry in shaping AI’s influence on education.

Note: As before, this post has been (mostly) crafted using ChatGPT (v4). I have modified and tweaked aspects of the prompt and output so (a) I understand it and the process better, and (b) it reads a little bit more like something I would have written, but it is mostly LLM-created.

The ethical considerations for prompt engineering touch on issues of bias, transparency, accountability, privacy, and beyond – something I’ve heard referred to as ‘Conscientious Commands‘ (I quite like that). As we look to understand and introduce AI to learning and educational technology, we need to be mindful of these considerations, understanding how to guide educators and technologists through the labyrinth of ethical AI communication to ensure that these digital assistants serve the greater good.

Bias:
Bias in AI is not just a technological issue; it’s a reflection of our societal inclinations that seep into our interactions with machines and the data the LLM has been trained on. When we construct prompts for AI, we’re often unaware of the prejudices embedded in our language or the tool itself. These biases can skew AI responses, leading to a reinforcement of stereotypes or the exclusion of underrepresented groups.

Expanded Context: For instance, when asking AI to provide common challenges faced by students, we must be wary of implying a homogenous student body. Educational experiences vary widely due to socioeconomic, cultural, and personal factors, and our prompts should reflect this diversity.

Deeper Reasoning: By meticulously crafting our prompts to be inclusive and representative of all learners, we can instruct AI to consider a broad range of experiences, which, in turn, helps to democratize the knowledge it provides and supports a more equitable educational environment.

Transparency and trust:
Trust is the cornerstone of any educational endeavour, and transparency in prompt engineering is a prerequisite for trust in AI. The prompts we engineer often dictate the AI’s line of reasoning, and if framed incorrectly, they can obscure the multifaceted nature of educational concepts, leading to oversimplified or biased information.

Expanded Context: When we ask AI to determine the “most effective” teaching method, we must recognise that effectiveness is subjective and dependent on context. Our prompts should reflect the complex, dynamic nature of pedagogical success and encourage AI to present a range of methods validated by different studies and expert opinions.

Deeper Reasoning: Creating prompts that elicit a spectrum of responses, rather than a single ‘correct’ answer, builds a foundation of trust with the users. It ensures that educators and students are receiving a well-rounded understanding of the topic, which is essential for informed decision-making and critical thinking.

Accountability:
The response AI gives can significantly influence learning outcomes and perceptions. Therefore, the responsibility of ensuring accurate and reliable information falls on the shoulders of those who engineer the prompts.

Expanded Context: When we direct AI to source credible information, we must be explicit about the standards of credibility we expect. This means crafting prompts that not only seek information but also prioritise the quality and reliability of that information.

Deeper Reasoning: The onus is on us to continuously refine our prompts and validate the AI’s responses, thereby maintaining a high standard of educational content. This accountability safeguards against the dissemination of misinformation and upholds the integrity of the educational resources provided by AI.

Privacy and personalisation:
The allure of personalisation in AI-driven education is undeniable, but it must be balanced with an unwavering commitment to privacy. As we tailor prompts to garner individualised or personalised responses, the sanctity of personal data and the rights of users must remain at the forefront.

Expanded Context: Prompting AI to analyse learning patterns necessitates consideration for consent, data minimisation, and anonymisation. We must ensure that the data used to personalise learning does not compromise student privacy or violate data protection laws.

Deeper Reasoning: The ethical engineering of prompts requires us to navigate the fine line between personalisation and privacy. By doing so, we uphold the trust placed in educational institutions and protect the rights of learners, ensuring that personalisation enhances rather than exploits the educational experience.

Preparing for the unintended:
AI’s literal nature can sometimes lead to unintended and undesirable outcomes. Ethical prompt engineering involves not only directing AI towards appropriate responses but also safeguarding against potential misuse of the information provided.

Expanded Context: A prompt requesting methods to maintain student attention could inadvertently lead to suggestions that infringe on privacy or autonomy if not carefully constructed. It is crucial to anticipate how AI might interpret a prompt and ensure that the boundaries of ethical educational practice are clearly communicated.

Deeper Reasoning: By contemplating the broader consequences of our prompts, we reinforce an ethical framework within which AI operates. This foresight prevents the possibility of promoting intrusive or unethical educational practices, thus preserving the integrity and humanistic values of education.

The exploration of ethical prompt engineering and ‘conscientious commands‘ serve as a reminder of the profound influence contained by the questions we pose to AI. As users of this technology, we must engage with AI with an acute awareness of the ethical dimensions our prompts encompass. By adhering to these principles, we ensure that our digital advancements in education remain respectful, equitable, and human-centred.

Photo by Emiliano Vittoriosi on Unsplash