Hi! For a project I’m working on, we asked GPT4 to design the structure of a simple Qualtrics survey and describe what it had done. With some hand-holding, it came up with a reasonable structure. On a whim, we asked GPT to create the corresponding QSF file for survey, which it did. We then tried to import it, producing a variety of file parsing errors. We went back and forth with GPT asking it to fix the errors or simplify the design, but none of the results were importable. 

I’m attaching one of the files here:

{
  "SurveyEntry": {
    "SurveyName": "Customer Satisfaction Survey",
    "SurveyDescription": "A survey to measure customer satisfaction with recent purchases."
  },
  "SurveyElements": n
    {
      "ElementType": "BL",
      "PrimaryAttribute": "Introduction",
      "SecondaryAttribute": "Thank you for choosing our products. We appreciate your feedback on your recent purchase.",
      "TertiaryAttribute": null,
      "Payload": {}
    },
    {
      "ElementType": "SQ",
      "PrimaryAttribute": "Introduction",
      "SecondaryAttribute": "Thank you for choosing our products. We appreciate your feedback on your recent purchase.",
      "TertiaryAttribute": null,
      "Payload": {}
    },
    {
      "ElementType": "SQ",
      "PrimaryAttribute": "What is your age range?",
      "SecondaryAttribute": null,
      "TertiaryAttribute": null,
      "Payload": {
        "Choices": {
          "1": {
            "Display": "Under 18"
          },
          "2": {
            "Display": "18-24"
          },
          "3": {
            "Display": "25-34"
          },
          "4": {
            "Display": "35-44"
          },
          "5": {
            "Display": "45-54"
          },
          "6": {
            "Display": "55-64"
          },
          "7": {
            "Display": "65 or older"
          }
        },
        "QuestionType": "MC",
        "Selector": "SAVR"
      }
    },
    {
      "ElementType": "SQ",
      "PrimaryAttribute": "How satisfied are you with your recent purchase?",
      "SecondaryAttribute": null,
      "TertiaryAttribute": null,
      "Payload": {
        "Choices": {
          "1": {
            "Display": "Very Unsatisfied"
          },
          "2": {
            "Display": "Unsatisfied"
          },
          "3": {
            "Display": "Neutral"
          },
          "4": {
            "Display": "Satisfied"
          },
          "5": {
            "Display": "Very Satisfied"
          }
        },
        "QuestionType": "MC",
        "Selector": "SAVR"
      }
    },
    {
      "ElementType": "SQ",
      "PrimaryAttribute": "Please provide any additional comments or feedback:",
      "SecondaryAttribute": null,
      "TertiaryAttribute": null,
      "Payload": {
        "QuestionType": "TE",
        "Selector": "ESTB"
      }
    },
    {
      "ElementType": "SQ",
      "PrimaryAttribute": "Closing",
      "SecondaryAttribute": "Thank you for your valuable feedback!",
      "TertiaryAttribute": null,
      "Payload": {}
    }
  ]
}

It appears that the file has a single block and simple questions, but that the questions should be listed in the block payload. It is also missing a survey flow element, notes elements, and so forth. 

I was wondering if someone could point me to a resource that describes the various elements, which ones are required, and other information that would enable me to either write a script to filter or augment the files that GPT produces or better yet, teach GPT how to improve the JSON it produces. The best resource I’ve found so far for this is: https://gist.github.com/ctesta01/d4255959dace01431fb90618d1e8c241

but this has only gotten me so far. Any help anyone could provide would be greatly appreciated.

Thanks!