Systems | Development | Analytics | API | Testing

Support for model fine-tuning in BigQuery takes input text (the prompt) and the expected output, and teaches the model to mimic the behavior.

The topic of testing AI and ensuring its responsibility, safety, and security has never been more urgent. Controversy and incidents of AI misuse have increased 26-fold since 2021, highlighting growing concerns. As users quickly find out, AI tools are not infallible; they can make mistakes, display overconfidence, and lack critical questioning. The reality of the market is that AI is prone to error. This is exactly why testing AI is crucial. But how do we test AI?

We all love the ease and affordability of SaaS (Software-as-a-Service) tools. We expect our SaaS apps to be available whenever we need them, building our business processes around this assumption. At WSO2, we are grateful that many organizations trust our SaaS solutions, and we continually work to ensure this trust is well-founded. It’s worth taking a moment to understand the impact when those mission-critical apps go offline. Just one minute of downtime can cost businesses thousands of dollars.

Extracting values from server responses is a common operation in scripting scenarios for performance testing. In most cases it is a mandatory step in order to reproduce realistic traffic using dynamic values generated on server side like sessionIds or Tokens. The most known way to extract values is using regular expression. While regular expressions are efficient for raw text-based responses, they are less efficient for structured responses like XML or JSON.

Marketers everywhere are anxious about what fortune the cookieless future holds — but I think we should embrace what will be a refreshing change.