FAQ – AI Frequently Asked Questions
GETTING STARTED
Is AI automatically included in Sphinx solutions?
Yes, AI is included in our solutions. A certain number of tokens is offered upon subscription. Beyond this limit, a subscription will be required. For more information, feel free to contact our sales team at contact@lesphinx.eu.
How do I enable AI on my account?
-
Click on the AI button for the corresponding feature.
-
Click on Go to the next step.
Acceptance of the general conditions specific to AI assistants is required.
-
Read the specific terms and conditions for AI Assistants carefully.
-
Check the following box:
-
"I acknowledge that I have accepted without reservation the ‘AI Assistant’ addendum displayed above and I declare that I am the legal representative of the holder of this account or have the authority to represent them validly and that I accept liability in the event of any false declaration".
-
-
Click on the I accept button.
-
GENERAL INFORMATION
Can AI make mistakes?
Yes, like any program, AI can produce errors. We always recommend reviewing the results before using them.
What is an LLM?
An LLM is a Large Language Model, a type of artificial intelligence program capable of recognizing and generating text, among other tasks.
USAGE & LIMITATIONS
Which AI model do you use?
At Sphinx, we use a model developed by ChatGPT. Upon request, it's also possible to use Mistral (which ensures data processing within Europe and guarantees zero data retention).
How can I improve the quality of AI responses?
For questionnaire generation, use clear and precise instructions. The more structured your prompt, the more relevant the response will be. For data interpretation, we recommend specifying the context and objective of your study in the dedicated fields. This context will be used for all automatic interpretations within that study.
Can AI analyze open-ended questions and verbatim responses?
Yes, AI can structure them, identify trends, and suggest thematic categorizations. This saves time when working with qualitative data, but human validation is always recommended, and manual adjustments can be made if needed.
PRIVACY & SECURITY
Who is the legal representative responsible for signing the AI usage consent agreement?
The legal representative is usually the head of your organization or anyone with the authority to legally represent it.
Does your AI model retain my data?
With the Mistral LLM, Sphinx has chosen a Zero Data Retention approach. This means that data is only stored for the time necessary to generate the content. For more details, please contact your sales representative.
Is my entire survey database sent when I use AI?
Only the selected variable data is sent in prompts to the LLM. In addition, our iQ3 survey tool will soon include features to anonymize text variables.
CREDITS
What is a token?
In the world of AI, a token (also called a credit or unit) is a basic unit used to break down and represent data, mostly text.
How can I estimate the subscription level I’ll need?
You can contact our sales team, who will provide estimates based on the type of content you want to generate.
Does running the same interpretation twice consume tokens twice?
Yes, running the same interpretation twice will result in double token usage, even if the output is identical. Each execution is counted separately.
Why do I get two different results if I generate a codebook twice? Is it a normalization issue?
It’s similar to two humans doing coding! The results can vary.
Is there a way to track my token consumption?
When launching an AI interpretation or coding, you’ll see an indicator showing the number of tokens used compared to your available balance. For more information, please contact your sales representative.
Will I be notified if I reach my AI token limit?
Yes, an alert message will appear if your credit balance is insufficient.

