Interdisciplinary Prompting in ChatGPT
One of my favourite ways to explore ideas for solving problems is using ChatGPT.
As someone who loves writing, planning and designing productive solutions, ChatGPT is a great assist for exploring ideas. One of the ways I use ChatGPT for problem-solving is using a technique I call 'Interdisciplinary Prompting'.
Interdisciplinary Prompting comes from ‘Interdisciplinary Thinking’, which I see as the mental habit of exploring and bringing inspirations from multiple disciplines to solve problems that are usually siloed into their default disciplines.
I was inspired by Nazam Laila , my former classmate at Brac University, whom I've known as a strong?advocate for interdisciplinary approaches in research. As a layperson, I took the idea of Interdisciplinary Research to ‘Interdisciplinary Thinking’ and explored how this approach could be used in ChatGPT for better prompting.
This post provides a brief introduction and demonstration of how we can use Interdisciplinary Thinking in combination with the well known Prompt Chaining technique to Prompt ChatGPT for exploring solutions to our problems.
How to do ‘Interdisciplinary Prompting’
The idea for ‘Interdisciplinary Prompting’ is simple. Instead of asking ChatGPT for direct answers, we take time to design a prompt that draws on other relevant disciplines. By Combining Interdisciplinary Thinking to design the prompt and then training ChatGPT using the Prompt Chaining technique we can generate a lay of the land and then expand on the ideas as we require.
And this is how it works:
Let's say we have a hypothetical problem. As a solopreneur, we are interested in exploring ideas on how to market ourselves. We have some idea of how it might look, but we want to explore more in depth. Usually, what we would do is ask ChatGPT directly about it. i.e.: 'Hey ChatGPT, how to do marketing as a Solopreneur?'
This should generate something like this:
Marketing as a Solopreneur:
(Note: I have used custom prompt parameters to keep the response short and simple.)
In 'Interdisciplinary Prompting' we take a more indirect approach. First, we identify the variables and then custom design a prompt with the variables we defined. The workflow could look like the following:
Now all we have to do now is create a custom prompt using the variables <the problem> & <the discipline to explore>. Example meta-structure could be:
<the problem>, and I want to know how <the discipline to explore> would approach this problem.
For example: “I want to learn Marketing Techniques and I want to know how Anthropology would approach this problem.”
Or
<the problem>, and how an expert in <the discipline to explore> would do it.
For example: “I want to learn Marketing Techniques, and how an expert in Anthropology would do it.”
领英推荐
We can also get creative and ask in natural language using the keywords appropriately. For example:
'What are some great wisdoms we can learn from Anthropology and use as techniques in Marketing?'
I got the following result for this prompt:
Wisdoms from Anthropology for Marketing:
We can see the differences right away. In Marketing ‘Defining an Audience’ or ‘Email Marketing’ is what most of us are already aware of. But things like ‘Understanding Cultural Context’ or ‘Emphasise Storytelling’ goes much beyond existing practices, and that adds depth to our explorations.
We can also make more variations such as using ‘Psychology’ or ‘History’ instead of Anthropology.?
Here is a result for Psychology:
Wisdoms from Psychology for Marketing:
And a result for History:
Wisdoms from History for Marketing:
Notice that we did not cram more than one discipline because it can degrade the quality of answers, especially if we are using lightweight models where token counts may be restricted. By modifying certain custom output settings and using models with better capacities, we can get some very satisfying, and in-depth answers from ChatGPT.
That being said, I encourage you to try and experiment with this technique yourself. Get creative, add in additional variable dimensions such as location, knowledge-sources, people, culture, language in unique combinations to generate interesting, in-depth and personalised results.
Final Thoughts
To sum up, Interdisciplinary Prompting is the practice of combining Interdisciplinary Thinking and Prompt Chaining to explore and ideate solutions for problems that are usually explored within its default?disciplines. By exploring other disciplines, it is possible to get some out-of-the-box results that are unique and not visible in usual prompting.
Of-course ChatGPT comes with the baggage of biases and hallucinations, so take the contents generated?with a grain of salt. What is exciting though is the subtext and the patterns. It's amazing how easily ChatGPT can tap into the corpus of knowledge and connect concepts between not just disciplines but also across history and geography. Once you discover this, using ChatGPT becomes much more fun to use.