The use of artificial intelligence by a company can be a gamechanger. But it also could impact employees and customers in ways that one doesn’t imagine. And it could be irreversible.
So, what should you do about onboarding AI software into your company?
- Assess your need: What is the problem you are trying to solve? This is NOT like a kid’s birthdays party goodie bag that you take because “why not” and throw in a junk drawer to purge later. The White House’s AI Bill of Rights requires this ex ante check. So does the NY AEDT Act, if you are using automated tools in the employment context; And so does GDPR, CPRA and the other U.S. privacy laws that require data minimization if personal data is involved.
- Understand the AI solution that you are onboarding: Beware of black-boxes. Like the nutrition label in your food, if you can’t pronounce the ingredients it’s probably bad for you). You should be suspicious if your vendor isn’t able or willing to explain the tech to you.
- Figure out whether or not the solution is fit for your purpose: Can it get the result it says it can? Is it reliable? It is important to understand the potential risks and how they can be mitigated. This is an AI Audit, and is required by the White House’s AI Bill of Rights, various privacy laws and the EU AI Act. There also are existing frameworks for this, like the National Institute of Standards and Technology (NIST) AI Framework.
- Provide transparency for the product that you are using and consider whether and how to provide human intervention. (This is one of the White House’s principles, but it also is part of the existing Colorado CPA rules the upcoming CPRA regulations.)
- Make sure that the guardrails that you determine are necessary are actually put in place. For that, you need to have a voice within the company, as well as allies and champions.