The State of Digital Quality in Functional Testing 2026

AI has become ubiquitous – especially within software development and testing. But as organizations integrate new tools and techniques into their workflows, growing pains abound. Teams struggle to adequately test massive volumes of code, determine which tasks are best left to human judgment, and create clear guidelines about when, where and how to use AI in development and QA. This report offers a snapshot of where organizations are on the journey to evolve their development and testing processes and crucial considerations for maintaining quality during this paradigm shift.

Change is happening fast. In Applause’s 2025 digital quality benchmark survey, 59.6% of respondents reported using AI in the testing process. Just over a year later, that number has risen to over 92%: Only 7.9% of respondents stated that they don’t use AI for any aspect of testing. Our most recent survey attempted to quantify AI’s impact on the way software engineering and testing work gets done.

Though AI appears to have a slightly larger impact on development than on QA, the margin is slim. The gap may narrow further as AI-powered test automation and LLM-as-judge gain wider adoption.

While many teams are turning to AI for test automation, self-healing automation can still be tricky according to Applause CTO Tacita Morway. ”When an automated test fails, it either just stops, or worse yet, an AI-powered system may optimize for completing the task – even if that means changing the test so it passes without actually checking the behavior it was supposed to test. Essentially, it changes its own rules. Safe self-healing automation has to understand the intent of the test, not just the automated steps. With that context, the system can adapt to legitimate changes in the application while avoiding false positives, hallucinations, or gaming its own system just to get to a passing result,” she said.

While using AI can create speed and efficiency for software teams, it also poses new challenges — some more readily apparent than others. Currently, the ability to create more code faster outpaces most organizations’ capacity to effectively test that code, introducing risk.

New Quality Challenges AI is Creating

Since introducing AI into the SDLC, only 26.4% of survey respondents have seen both the number and severity of defects reaching production decrease. Some teams reported no change (and 19.8% said they don’t track that data, which is another problem), while others face higher numbers of issues, more severe issues, or more often, a combination of the two.

Human involvement in functional testing remains extremely important for 86.1% of respondents, with an additional 13.4% considering it somewhat important. Fewer than 1% believe human involvement is not at all important in functional testing.

While AI has been touted as a force multiplier, it’s not a magic bullet. To get the most benefit, software organizations must find ways to create more intelligent, efficient testing methodologies to validate the tsunamis of code AI can generate, while incorporating human judgment at critical points to reduce risk and create better, smarter QA.

Report Methodology

In August 2026, Applause conducted a survey of members of the uTest community as well as other software development, QA, product, AI and data science professionals, with the following goals:

• Identify where they are incorporating AI in their software development and QA processes
• Quantify the impact of AI on digital quality
• Understand key challenges with integrating AI into the SDLC

We also conducted interviews with technology leaders.