On May 8, 2024, Colorado become one of the first states to begin to enact a law which prohibits employers from using artificial intelligence (AI) to discriminate against their workers. This new law would require companies to take extensive measures to avoid “algorithmic discrimination.” Algorithmic discrimination are errors that are present in computer systems which lead to people of protected classes being inadvertently discriminated. If the law is enacted, then it would go into effect in 2026 and would specifically target discrimination that arises from AI being used to make important decisions such as hiring and firing.

What does this mean for employers and who does it apply to?

This bill, if it is signed by the governor of Colorado, would only apply to companies that develop or use “high-risk AI systems.” High risk AI systems are those which are used to make substantial and important decisions such as education enrollment, employment opportunities, financial lending services, housing, insurance and legal services. When companies use high risk AI systems, it could lead to algorithmic discrimination such as protected classes not getting access to housing at the same rates as un-protected classes, people with disabilities being fired or not hired at higher rates than people who do not have disabilities, or protected classes of people getting higher interest rates on loans than other people.

The bill would apply to developers and deployers of high-risk AI systems and would put a legal duty on them to use reasonable care to avoid algorithmic discrimination. Developers are any person that develops or intentionally and substantially modifies an AI system. Deployers are any person that uses a high-risk AI system.

Developers, under the bill, will now be required to make information available to deployers and other users summaries of the type of data that is used to train high-risk AI. They would also be required to make a public statement which describes how the developers manage known or reasonably foreseeable risks of algorithmic discrimination.

Deployers will have some requirements and expectations as well. The bill would require deployers to implement risk management policies which govern their internal use of high-risk AI, regularly review this policy, complete assessments of the AI system annually and most importantly, notify consumers when high-risk AI will be used to make a consequential decision.

Conclusion

While this bill is the first of its kind to have been drafted and created in the nation, other states will soon follow their lead. California and New York are currently debating and discussing similar laws in their state legislature. As mentioned above, the bill still needs to be signed and approved by the governor before it is fully enacted however, this bill is a step in a right direction for AI to become more mainstream. For those who were skeptical in using AI, bills like this one might provide some assurance that AI will be monitored and closely regulated. For those unsure on how to use AI, this bill and similar ones that are adopted across the nation, might propel people to start to use AI more because they finally have the guidance they need. Overall, this bill signals that whether we like it or not, AI is here to stay.