AI Discrimination Policy
Ometrics LLC
Last Revised January 1, 2026
Summary
Ometrics AI and Ochatbot, from here on called “Ometrics AI” or “the AI” is presented as a chatbot icon and or a search field on a web page. The AI uses data from the website and or data provided by the client as source data to learn from. There are two types of AI, Scripted AI and Generative AI. In some cases, there is logic implemented for decisions before generative AI is activated.
Different LLM’s are used to interpret what the user is asking and to translate the result back to the user in the language the user is communicating in. The generative AI is a RAG system. This means only knowledge gained from the data provided is used, not knowledge in the LLM training data set.
Limitations
The AI is limited by the data. The knowledge base of the AI is based on the website data and any added data. If a question is asked and there is no data about that question, a Fallback is triggered, and the user is asked to contact the owner of the site, switch to live chat, fill out a form as a ticket, or call to have their question answered.
Purpose
Ometrics AI can be used for an unlimited number of use cases. Typically, it is used in 1) ecommerce systems to recommend products and handle customer service questions after a purchase. 2) lead generation to overcome an objections, answer questions, and offer the customer a form to get a quote or more information. 3) support for a company’s customers, answering questions, and when not able to, creating a ticket which is sent to the clients ticketing system.
Intent and Outputs
Ometrics AI is used to answer questions accurately and provide 1) options for a purchase 2) an answer to a question 3) a way to send information to the company.
Consumers are aware they are interfacing with AI because the interface says it’s powered by AI. We also recommend to our clients that the engagement question also states that this is an artificial intelligence, not a human. That said, it is very obvious that they are interfacing with AI, not a person, by how the AI responds and shows products, blogs and other data.
AI Evaluation, Data Governance, and Procedures
Ometrics AI is tested on a regular basis. When system-wide updates are made they are posted in the monthly update newsletter to all clients. With each new feature, there are two different testing groups evaluating the change. Active AI systems are tested via This is a Human-in-the-Loop.
- For DIY clients, they are responsible to check their fallbacks and conversation data to make sure questions are being answered and answered correctly and stating that the interface is AI.
- For clients paying Ometrics to manage their AI an Ometrics employee is testing the AI on a scheduled basis. Ometrics employees report back to the client any issues with their data as well as insights to help their business.
- All clients are responsible for the data on their website and the data they add to the AI. If they create prompts in the data set, provide false information for the AI to learn from, it is the client’s responsibility. If Ometrics learns or detects data that is discriminatory, the client must update the data or face having their account closed.
If Ometrics detects an unknown algorithmic discrimination that has caused harm or persuaded a decision in a biased manner, Ometrics employee, client, or client’s consumer who discovered the issue will report it to Ometrics Compliance or simply use our contact us form. Then Ometrics will review and fix the issue, and will notify the Attorney General of Colorado of the issue and the solution and or action taken. If consumers were affected, appropriate notification will be sent by Ometrics or the Client. Note that it is not possible to tell who interacted with the AI, but those who purchased or filled out a form would be notified.
If you have any questions about this policy, the practices of this site, or your dealings with this website, please contact us at [email protected], or send mail to:
Ometrics LLC
PO box 2095, Evergreen CO 80437