Charles Ungerleider, Professor Emeritus, The University of British Columbia
[permission to
reproduce granted if authorship is acknowledged]
The development of large language models
like ChatGPT and their use in education and other fields are occurring so
rapidly that it is difficult to keep track. Teachers play an important role in
showing students how to use them well.
Teaching students to formulate prompts to
obtain useful and accurate responses will serve the students well in using
large language models or making inquiries of other sources. My advice to
students would be to:
- use simple language that is phrased clearly to avoid ambiguities or irrelevances.
- include relevant background information to contextualize the question.
- be specific, narrow the question or prompt.
- phrase the prompt as a direct question.
- break complex topics into smaller, manageable parts.
- determine the desired format or structure of the response.
- limit the length of the response.
- avoid biasing the response.
- use examples or scenarios to focus the response.
I’d want them to know that prompts or
questions can be designed to gather specific types of information or insights:
Descriptive questions seek detailed
information or descriptions of situations, events, or concepts. Example:
"Describe the main features and functions of a smartphone?"
Classification questions seek to categorize,
sort, or classify things based on their attributes, properties, or
relationships. Example: "What are the main categories of programming
languages, and what are their distinguishing characteristics?"
Inferential questions require drawing
conclusions, making inferences, or deducing information based on available data
or premises. Example: "Given increasing global temperature trends,
what can we infer about the potential impacts of climate change on sea
levels?"
Comparative questions focus on comparing
different entities, situations, or ideas to highlight similarities and
differences. Example: "What are the similarities and differences between
solar and wind energy in terms of efficiency, cost, and environmental
impact?"
Evaluative questions seek opinions, judgments, or assessments about a subject, often involving criteria or standards. Example: "In your opinion, what are the most desirable attributes of a student body president, and why?"
Hypothetical questions explore potential
scenarios, situations, or outcomes to encourage creative thinking and
problem-solving. Example: "If you could develop a new technology to
address a major global issue, what would it be, and how would it work?"
Causal questions seek to identify causes,
effects, or relationships between different factors or events. Example:
"What are the primary causes of deforestation, and how does it impact
biodiversity?"
Problem-solving questions seek solutions,
strategies, or advice for addressing specific issues or challenges. Example:
"Are there steps that this school could take to reduce online
bullying?"
Reflective questions encourage
introspection, self-awareness, and personal growth by examining one's thoughts,
feelings, or experiences. Example: "What have you learned from a past
failure, and how has it influenced your approach to similar situations in the
future?"
By teaching students how to form good
prompts, understand the different purposes of questions, and craft well-formed
prompts or queries, teachers can help students sharpen their inquiry skills and
use sources such as large language models like ChatGPT productively and
meaningfully.