UPCOMING WEBINAR: DIY Data Cleaning & Prep: Super Fast, Super Good AI presents a nearly endless list of possibilities when it comes to working with data. Displayr has integrated AI in a number of ways, including assisting in text categorization and using prompts to build out your analysis. This article lists some of the most frequently asked questions (FAQs) about using AI in Displayr. For a more detailed review of the available options in Displayr, see Get Started Using AI in Displayr.
- What is Displayr doing to incorporate AI into the software?
- What data gets processed by AI?
- Where does my data get processed?
- Can I choose which AI model to connect Displayr to?
- Are AI functions included with my Displayr license?
- How do I use my OpenAI account in Displayr?
- How do I opt in or out of AI in Displayr?
- What models does Displayr use for AI?
- Are my AI prompts shared with anyone else or used to train the AI?
- How do I use AI in Displayr?
- Can AI analyze my data?
- Can AI interpret my data?
- Can AI interpret a table that contains significant differences (e.g., arrows and font colors)?
- How do I use AI to summarize my findings?
- Can I use AI to generate images?
- Can AI automate data cleaning and preprocessing?
- Can AI help find patterns or trends?
- How accurate is AI in making predictions from data?
- How can AI be used to categorize open-ended text responses?
- Can I extract tables or statistical analysis into an R or JSON file to use in a large language model?
What is Displayr doing to incorporate AI into the software?
Displayr has integrated AI in a number of ways. You can use Displayr AI to help with text categorization and label tidying. To leverage AI to help you with data summarization and interpretation, as well as image creation, other text categorization tools, and help with code, you will need to connect your OpenAI account. See Using OpenAI in Displayr for ways to use these features.
What data gets processed by AI?
When using any AI function, only the data you provide to the function is included in the prompt to the AI (regardless of which type). For example, when conducting text classification, all responses and metadata for the question(s) being classified are included, but none of the other data in your data file or document.
Where does my data get processed?
Displayr currently connects to multiple AI solutions for different features:
- Text classification (categorization) and summarization of text data will be processed in the locations outlined here: Displayr AI Processing Location.
- Any feature that requires OpenAI credits will be processed on OpenAI's servers, which are primarily located in the United States of America. See OpenAI Platform - Infrastructure
- The Displayr AI Assitant processes data in the United States of America. See Alhena Terms of Service
Can I choose which AI model to connect Displayr to?
This is not currently possible when using any of Displayr's pre-built features. We are working on a code-based (R) solution that will provide better control of this in the future.
Are AI functions included with my Displayr license?
The Displayr AI-assisted text categorization and label tidying features are included with your Displayr license. If you'd like to use AI functions that include a customizable prompt (i.e., create images, interpret outputs, summarize text, etc.), then you will need an OpenAI account with API credits. Note that OpenAI credits and a paid ChatGPT account are two different products. A paid ChatGPT account cannot be used to access OpenAI through Displayr. OpenAI API credits are purchased separately. See the OpenAI website for details about pricing and to sign up.
How do I use my OpenAI account in Displayr?
Once you have an OpenAI account with API credits set up, you will need to connect Displayr to OpenAI using the steps found in Connecting Displayr to OpenAI.
How do I opt in or out of Displayr AI?
You can opt in or out of Displayr AI via your account settings. Click on the initials icon in the upper right corner in Displayr and click Account Settings. Scroll to the Displayr AI section of the General tab and toggle the dropdown between Enabled and Disabled, depending on what you want to do.
If you are updating the dropdown to Enabled, then you will be prompted to view and agree to the Displayr AI terms. If you do not agree to the terms, then Displayr AI will remain disabled.
What models does Displayr use for Displayr AI?
Displayr AI uses AI models on the Azure OpenAI to process and generate data. Displayr AI uses fine tuned models that are trained by Displayr Data Scientists on non-customer data to better categorize text responses and label data.
Are my AI prompts shared with anyone else or used to train the AI?
The prompts and data are not shared with anyone or used in training. The prompt is constructed within the Displayr software and sent to the AI model on Azure OpenAI. It is processed by the server and the result is sent back and displayed in Displayr. See the Azure OpenAI data, privacy, and security page for more detail.
How do I use AI in Displayr?
AI in Displayr can be accessed in a variety of ways, depending on the actions you'd like to use AI for.
If you'd like to use Displayr AI for text categorization see How to Classify Text Data. If you're looking to tidy labels on import or when combining variables, this will happen automatically if you have Displayr AI-enabled.
If you're looking to create stand-alone AI outputs using prompts, see Using OpenAI in Displayr.
Can AI analyze my data?
You can use AI to interpret outputs, perform data analysis on variables (create a crosstab, for example), and summarize text outputs.
To perform actual calculations for analysis, it is recommended to use our Data Analysis feature. In the example below, I can feed two different variables (gender and age) into the AI > Data Analysis inputs and prompt it to create a crosstab:
Can AI interpret my data?
You can use the AI > Interpret function to interpret outputs such as a table or regression.
For example, I have a crosstab that shows different age groups' interest in visiting a list of countries:
The table is a bit large and it can be hard to pull out quick insights. I can use the AI > Interpret tool and use the crosstab as the input and provide a prompt asking for a summary in 100 words or less:
Can AI interpret a table that contains significant differences (e.g., arrows and font colors)?
If you have turned on arrows and font colors or compare columns on a table, you can use AI > Interpret and the significant results will be included in the AI-generated interpretation summary. The example above includes the significant differences in the interpreted summary.
How do I use AI to summarize my findings?
You can use AI to summarize text outputs/callouts in your Displayr project. For example, I created a report about a healthy burger concept and have a few short summaries for each table:
I can use AI > Summary and use the text outputs associated with each table to create a high-level executive summary:
Can I use AI to generate images?
Yes, go to AI > Image and then enter a prompt for your image.
Can AI automate data cleaning and preprocessing?
Currently, you can use AI in Displayr to clean text variables in your data set. You can also use AI to translate and/or classify text data.
Can AI help find patterns or trends?
You can use the Custom prompt to summarize patterns and trends in selected variables. For example, you can select a text variable and create a prompt to read all text responses and return a summary of the top 3 themes in the responses.
How accurate is AI in making predictions from data?
Results from AI features within Displayr may contain material inaccuracies and may not reflect correct, current, or complete information. Do not rely on or encourage others to rely on any results without independently evaluating their accuracy and appropriateness of use, including by using human review. Displayr makes no representations or warranties and provides no guarantees regarding the accuracy of results.
How can AI be used to categorize open-ended text responses?
There are two different ways that you can use AI in Displayr to categorize open-ended responses.
If you want more control over AI-generated themes and how text responses are assigned to themes see How to Classify Text Data.
If you want a quick and dirty categorization, select a text variable in the Data Sources tree, then click + > AI > Text Categorization. From there, you can edit and refine the Prompt to help create categories. If you use this method, you will not be able to reassign text responses to different categories, reuse the code frame, or update your data set and get the same results.
Can I extract tables or statistical analysis into an R or JSON file to use in a large language model?
This is currently not supported.