Have you received user feedback regarding the AI features, and how are you incorporating it into product improvements?

Getting user feedback is a very important element of building any AI feature. Without the user input, you cannot improve your AI, which means you cannot improve your product. Treat the AI feature similar to any product feature, you need to be nimble, agile and adaptable. There are a number of ways you can gather this feedback and depending on that you can incorporate it in many different ways

• In feature feedback: Think about this as the thumbs up or down you do on your chatGPT or ask it to re-generate. This is one type of user input to feed back to the algorithms that the output produced was not satisfactory to the user. This is human in the loop

  • I have used this when I was building chatbots 7 years ago by providing 5 emojis at the end of generating a response to the question. The ones that are on lower satisfaction, we would go analyze each of the answers manually to see what went wrong.

• Feature on/off: Lets say if you have an Opt-in / Opt-out feature, tracking the number of people option out can be a good way to know whether AI is working or not.

  • If customers are not signing up for the feature, there can be an awareness and onboarding problem. This can be fixed with some education material in the product as well as branding and messaging outside.

• Qualitative feedback - This is sitting down with your customers and showing them the AI and gathering input. This method us true of any product feature, even more important in the AI feature

Yes, we incorporate user feedback to improve the model and reduce drift over time. There are several ways to gather feedback:

1. Direct user feedback through thumbs-up and thumbs-down ratings.

2. User override rate – measuring how often users override AI-generated suggestions.

3. Tracking drop-off points where users disengage from the AI feature.

4. User support tickets – identifying recurring issues raised by users.

To incorporate feedback, we analyze failure points to determine the root cause. Common ways to improve the model include optimizing confidence thresholds, retraining with better data, and filtering out edge cases where responses are inaccurate or undesired. Additionally, the issue may not be with the model itself but with UI/UX design, transparency, or explainability—in which case, we focus on improving user experience and trust.