Can you share a practical example of how AI has been successfully implemented in your organization, and the impact it has had?

I frequently use GPT-4 to write code. It's a simple process that doesn't require an API. I use OpenAI and prompt engineering to get the AI to write programs for me. It's a significant time-saver. Sometimes, it completes the task entirely. Other times, it gets me about 75% of the way there, and I have to make some adjustments. It's written documentation for me and helped with data collection instruments. From a product standpoint, I've found the Bard models to be very useful for classification. I use these models to help build data structures and datasets. For example, I can use the model to classify narratives into different themes for thematic analysis. This has been particularly effective for a client and for internal use.

We have a product that's currently being piloted in a few public sector units in India. It's a meeting management solution that uses AI to record meetings, transcribe them, and even identify each speaker using voice biometrics. It generates complete meeting minutes, including a summary, key points discussed, and action items. It then assigns these tasks to the right people and follows up until they're completed. The product is still in its pilot stages, but we're hopeful for its potential.

There are several tools on the market that people are using and testing. Firefly, Otter, and Zoom all have similar features. Some of our partners have also tried out summarization tools and conversational bots internally. These aren't traditional chatbots; they're bots that leverage large language models. Initially, these were only used internally to avoid any potential reputational risks with external clients. But as confidence in the system grew, they began external trials to support the customer service team. These conversational bots have been quite successful and have resulted in high customer satisfaction rates, not only for customers but also for the human agents who would otherwise be overwhelmed with calls.