Why Utilizing Feedback Helps
Because artificial intelligence and large language models are so new in the realm of general use for the public, and more so as specialized platforms, they need a bit of... encouragement. Providing feedback helps reinforce the behavior of our platform, training it to do what is expected and avoid what is unintentional.
As we continue to build our products alongside our community of users, it truly takes a village to lend a hand in ensuring TaxGPT is the best it can be. Higher quality brings more exceptional value to everyone!
How to Use the Feedback Mechanism
As you interact with TaxGPT, you'll notice the feedback options in the form of Thumbs Up and Down icons:
If you click on either the thumbs up or down icon, you'll be prompted to provide feedback details. Simply type your feedback into the box and click "Send feedback":
When to Use the Feedback Mechanism
We really appreciate both good and bad feedback, as they are extremely valuable to the TaxGPT team as we continue to improve and reinforce our products.
Good feedback can be used any time the response is congruent with your expectations, in terms of content, sources, and formatting. Good feedback is especially useful when the product exceeds your expectations. For example, you may be asking about Research & Development credit information for a company in California, and not only does TaxGPT give you great information about that specific ask, but also gives you additional information regarding other business credits that may be applicable to the company in question.
Bad feedback is extremely valuable to our team and helps us pinpoints gaps in our product. Specific examples of when bad feedback should be used: hallucinations (incorrect information presented as the correct answer), lack of jurisdiction and/or tax coverage (i.e. Austin, TX Hotel Occupancy Tax), formatting issues, and incorrect or lack of sources.