Guidance for Appropriate AI Usage
Artificial intelligence represents a powerful advancement which could make our lives and jobs easier and more efficient. However, many of the risks remain unknown.
Amid all the noise and conversation regarding AI these days, it can be challenging to understand what the potential risks the various AI agents pose to Atomic Data’s clients. This post is due to an initiative by Atomic Data to provide guidance for clients on safe AI usage.
When using AI there are no assurances in place to protect the confidentiality or privacy of the data or information you use to interact with it. Any information you provide could be reused, repurposed, and become publicly available. It’s likely that giving non-public materials to an AI would be a breach in company policy.
Artificial Intelligence “creates” by relying on and re-using information which may be protected by copyright and other Intellectual Property laws. Even if it seems original, any output from an AI could be an infringement of intellectual property rights and could expose your company to legal action.
AI output can be unpredictable. Even when used appropriately, information obtained from AI may be outdated or factually incorrect and lack the proper citations to verify its authenticity. Instances of this are well-documented and numerous. Furthermore, biases in the information used to train AI can be reflected in the AI’s responses. All AI output should be reviewed by a human knowledgeable in the subject area before being relied upon.