Accenture launches AI-testing services to monitor online systems

On Tuesday, Accenture announced the introduction of new testing services for global Artificial Intelligence (AI) systems. These services are based on a ‘Teach and Test’ methodology, aimed at assisting companies in constructing, monitoring, and evaluating reliable AI systems either within their own infrastructure or in the Cloud.

Bhaskar Ghosh, Group Chief Executive of Accenture Technology Services, emphasized the importance of finding improved methods to securely and qualitatively train and sustain AI systems as organizations increasingly adopt AI. This approach is crucial to prevent adverse effects on business performance, brand reputation, compliance, and humans.

The ‘Teach and Test’ methodology by Accenture ensures that AI systems make accurate decisions through two distinct phases. The ‘Teach’ phase focuses on selecting the right data, models, and algorithms for training Machine Learning (ML). In the subsequent ‘Test’ phase, the outputs of the AI system are compared against key performance indicators, assessing its ability to explain how a decision or outcome was determined. This evaluation involves the use of innovative techniques and Cloud-based tools to monitor the system.

Kishore Durg, Senior Managing Director for Growth and Strategy, and Global Testing Services Lead for Accenture, highlighted the need for new capabilities in evaluating data and learning models, selecting algorithms, and monitoring for bias, ethical considerations, and regulatory compliance. Accenture’s ‘Teach and Test’ methodology addresses these concerns, providing companies with a comprehensive approach to developing and validating AI systems with confidence.

Accenture successfully applied this methodology to train a conversational virtual agent for a financial services company’s website. The agent achieved an 85% accuracy rate on customer recommendations and was trained 80% faster than previous methods allowed.