Prompts // 4 (Common Challenges)

My last three posts have covered aspects of using ChatGPT, as a demonstration of recent developments in AI and it’s power, to bring a literary classic up to date, I wrote about how to craft a prompt, and lastly about advanced prompting techniques and real-world applications of advanced prompting.

In this fourth post, I’ll cover aspects of the potential challenges and possible or realistic solutions.

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.

From the perspective of a robust understanding and strategic approach to the unique set of challenges AI offers us, we can look to overcome these hurdles to foster a meaningful interaction with AI. Below I’ll look at common challenges and propose detailed solutions:

Ambiguity in Prompts:
Challenge: Ambiguity in language can lead to misinterpretations, resulting in less accurate or irrelevant responses from AI.

Solution: Carefully crafting specific prompts is key: it’s advisable to avoid vague terms or conflicting terminology and ensure the context is well-defined.

Example: Instead of a vague prompt like, “What is the impact of technology?”, specify the domain and desired outcome, e.g., “What is the impact of educational technology on student engagement in higher education?”

  • Define Terms: Clearly define any terms that may have multiple meanings.
  • Provide Context: Offer a brief background or context to guide the AI towards the desired response.
  • Be Explicit: Specify the type of information you are seeking – whether it’s examples, statistics, theories, etc.

Lack of Contextual Understanding:
Challenge: AI may struggle with understanding deep or nuanced contextual information which could lead to off-mark responses.

Solution: Incorporate sufficient context in your prompt or employ follow-up prompts to clarify or provide additional information.

Example: If inquiring about a specific educational theory, provide a brief explanation or context about the theory in your prompt, or why you’re asking the question.

  • Follow-up Queries: If the initial response lacks depth, use follow-up queries to gather more information or clarify.
  • Provide Background Information: Offer relevant background information within the prompt to guide the AI’s understanding.

Overly Complex Queries:
Challenge: Multi-part or overly complex queries can overwhelm the AI or lead to incomplete or superficial answers.

Solution: Breaking down complex queries into smaller, more manageable prompts can yield more comprehensive responses.

Example: Instead of asking for a comparison of multiple learning theories in one go, break it down into individual queries about each theory, and then seek a comparison.

  • Incremental Prompting: Utilise a step-by-step approach to build up to the desired answer.
  • Separate Queries: Split multi-part queries into separate prompts to ensure each aspect is thoroughly addressed.

Misinterpretation of Terms:
Challenge: Terms with multiple meanings or industry-specific jargon might be misinterpreted or misused by AI.

Solution: Define crucial terms within your prompt to ensure correct interpretation.

Example: If asking about “engagement” in an educational setting, specify what aspects of engagement you are referring to, such as student participation, involvement, or attention.

  • Define Key Terms: Explicitly define terms that are crucial for the accurate interpretation of the prompt.
  • Avoid Jargon: Unless necessary, avoid using jargon or provide clear definitions to ensure accurate responses.

Expecting Interpretative or Subjective Answers:
Challenge: AI may struggle with interpretative or subjective queries that usually require human insight.

Solution: Frame your prompt to seek factual information or various perspectives that can inform your understanding.

Example: Instead of asking for the “best” practices, ask for “commonly recommended” practices or “research-backed” practices in a particular educational domain.

  • Seek Objective Information: Aim for prompts that solicit factual, objective information or a range of perspectives.
  • Specify Desired Format: If looking for interpretative insights, specify the format you’d like the answer in, such as summarizing various perspectives or providing a pros and cons list.

Through a detailed understanding of the challenge we face in using tools like ChatGPT, and the strategic solutions we may ask it for help with, we can significantly enhance our interaction with it, leading to a more insightful and actionable response.

Photo by Emiliano Vittoriosi on Unsplash