Prompt Engineering Tactics on Biological Intelligence
Here are some prompting tactics in case you want to practice prompt engineering on biological intelligence (known as
What questions do you have?Instead of
Do you have any questionsfor encouraging the human to ask questions. The first prompt assumes there are questions by default; the second doesn’t. You will likely get more questions with the second prompt.
What made you do it?Instead of
Why did you do itto have the human model answer the question without getting defensive. Why questions could unintentionally trigger a defensive response in a human.
- When arguing with a human, don’t tell them a mistake they made but focus on how a certain action made you feel to make a case that is easier to agree with.
- When pointing out a probable mistake, don’t declare
this is wrongbut rather say
I think this is wrongand include your explanation to create space for discussion.
- When dealing with a group of indecisive humans, make a clearly suboptimal recommendation to force the group to develop a better recommendation.
- Instead of asking questions that can be answered with a simple
no, use open-ended questions that encourage further elaboration and discussion.
Note that these prompts are tailored for specific versions of humans. For example, asking
would you like to do this will likely work better on a human of
Canadian version to get them to do something, whereas the same prompt could be stated as
do this for a human of the
German version for the same effect. You can therefore achieve the same results with fewer tokens when interfacing with different versions. Always test your prompts to see what works best!