Test Data Pro AI FAQ

This topic provides a concise overview of significant questions and answers concerning the utilization of artificial intelligence (AI) in the Test Data Pro features. For Perforce's AI policy, see Generative AI Policy.

What type of AI is being utilized for test data generation?

BlazeMeter utilizes Microsoft Cognitive Services. The AI-driven features employ OpenAI's services.

How can I determine if AI features are enabled or disabled? Whom should I contact?

For inquiries concerning account-wide AI consent settings, please consult your organization's BlazeMeter account administrator.

When you start the Data Creation Wizard, a global toggle element indicates whether the AI features are enabled or not. If your Admin disables the AI features, the Data Profiler and Test Data Creator will still work, but only the standard rule-based functions will be available.

How do I opt out of or opt in to these AI-driven features?

Your BlazeMeter Account administrators determine whether the account opts in or opts out in the global Account Settings.

What type of uploaded data does Perforce store?

We store all data in your test data model within your workspace, which you have full control over: CSV files are stored as individual files, and function parameters are stored as test parameters.

What are the AI features for test data generation?

The BlazeMeter test data integration has the capability to provide more accurate data profiling, as well as a wider range of generated data. Please note that OpenAI usage conditions are applicable, implying that test data generated by the AI service might include inaccuracies or unintended content beyond our control.

What happens if I choose to opt out later? Will tests that depend on the AI-driven Data Creator functions suddenly cease to function?

No, these tests will not stop working after opting out. AI functions tied to user consent are only utilized when generating test data, and the results are cached. That's the extent of it.

Additionally, there are other AI functions that don't require user data, and therefore, they will consistently function with your tests. Disabling the AI features will not impact the non-AI parts of the Data Creation Wizard.

Do these features require additional setup, training, or configuration?

No, the user interface will guide you. User training for Test Data Pro is available, but not required.

Could the test data generated by AI be too 'real'?

The generated test data is plausible and in a valid format, yet fictional. On the surface, the test data generators offer authentic-seeming data, such as real city names, ZIP codes, first names, last names, banks with RSSD IDs, and more. For instance, it can produce data that resembles actual credit card numbers or Social Security Numbers (SSNs) by employing the publicly-known formulas.

However, at a granular level, the generated test data is fabricated in the sense that it does not consist of personally identifiable information (PII). Random names, credit card numbers, addresses, phone numbers, SSNs, and the like, are independently generated and are not linked to any specific individual. Any resemblance to real persons is purely coincidental.

What are the best practices concerning confidential test data?

Only submit anonymized data for profiling. For testing, utilize exclusively anonymized or synthetic data. Avoid employing production data or any content that comprises confidential or personally identifiable information (PII) as test data. This guideline extends to CSV files and sample data uploaded to the Data Profiler and Test Data Creator as well: Refrain from uploading files that include confidential data or PII.

Does the Data Creation Wizard return different output when AI is enabled and when not?

If you cancel the AI enablement, the Data Profiler and Test Data Creator will still work, and only their built-in rules-based suggestions will be activated.

What are the limits (number of rows)? How long does it take to generate large amounts of AI-driven test data?

There is no limit to the number of rows. There is only a time limit of 5 minutes. You can generate as large an amount as can be transmitted from the AI service in 5 minutes.

Does BlazeMeter send a request for each data row? How long are the AI-generated values cached? How many data rows are cached?

We do not send a request for each individual data row. Rather, we retrieve 10 rows per each request. AI-generated values are not subject to caching. If necessary, we directly save these values into the test data model.

Can I utilize Data Parameters in AI functions like anySeedlist() and similarValues()?

Yes.

Can we customize or fine-tune the ML (Machine Learning) models for our specific needs?

No, the models are pretrained.