Skip to main content

Search

ai prompt tag and ai context tag

  • Updated

The ai prompt tag enables surveys to use AI to query a Large Language Model (LLM) and store the results in hidden questions. The ai context tag can be used to provide context that helps the LLM generate a better response.

In the example above, Q1X will categorize the respondent into one of the three options based on the open-ended response in Q1.

Details

  • The ai prompt and ai context tags give results in real-time. The tags can be placed on type: coded single select, type: coded multiple select, and type: coded text questions.
    • For closed-ended questions, the AI response(s) should match one or more of the available options, as shown above.
    • For open-ended questions, the AI generated text will be saved as the response.
  • If any conditional logic is applied, only qualified options will be used. 
  • The ai context tag is optional and references one or more earlier questions (open-ended or closed-ended) to provide relevant background based on respondent answers.
  • The responses generated can be used in survey logic such as branching or follow-up questions.
  • If there is no response after 20 seconds or if there are two failed connection attempts, the system applies default settings to ensure the respondent can still proceed past the survey page.
    • For closed-ended questions, option 99 will be selected if it exists, otherwise nothing will be assigned.
    • For open-ended questions, the value will be set to "n/a."

Note: The new type: coded text tag is designed to only be used with ai prompt and ai context.

Additional examples

Using with a multiple select question

In the example below, Q2X will select as many options as are appropriate based on the open-ended response in Q2.

Conditioning a follow-up question based on AI response

In the example below, Q3X will select as many options as are appropriate based on the open-ended response in Q3, and then Q3A will be asked if Q3X has any of the specified options 11 through 13 selected.

Creating a follow-up question based on an open-ended question

In the example below, Q4X will create a follow-up, open-ended question based on the open-ended response in Q4. Q4A will then be asked using that question if the conditions are met, in this case if that the question length is greater than three characters, indicating that it is not "n/a."

Creating a follow-up question based on a closed-ended question

In the example below, Q5X will create a follow-up, open-ended question based on the closed-ended response in Q5. Q5A will then be asked using that question.

Limiting the qualifying options

In the example below, Q6X will select as many options as are appropriate based on the responses to Q6 and Q6A, but will limit the available options based on the condition decorators.

Related articles

Comments

0 comments

Please sign in to leave a comment.