• Artificial intelligence (AI) is not an infallible process that resides on the computing equivalent of Mt. Olympus. AI results can contain bias or be misinterpreted, and companies need to put guardrails in place to prevent misuse.
• Microsoft decided to remove several products that could be used to manipulate information negatively. This will improve trust and accountability in the use of AI.
AI has already made a positive impact in a wide variety of fields, including businesses, science, and financial markets. Eventually, AI’s impact will influence every facet of daily life by increasing factory efficiency, solving complex problems, and improving the overall quality of life. However, in order to realize this future state, it is essential that the AI software that will control humanity’s critical systems be deployed in a manner that will impact life in a positive manner.
Recently Microsoft announced an update to its AI policy at its annual Build conference. Known as the Responsible AI Standard, v2, General Requirements, this update brings more specifics to the Microsoft policy, including better guidance on managing AI. Several AI tools were eliminated, including age and gender analysis tools that were designed to help persons with poor vision detect visual images and human emotions. Microsoft decided that although the tools could be helpful for the visually impaired, the company decided against releasing the application, given the difficulty in accurately interpreting emotional responses. Microsoft will also put limits on the use of its facial recognition software, which can be used to perform identity checks. Developers who want to use Microsoft’s facial recognition software must apply for access and explain how they plan to deploy it. Similarly, Microsoft will also require those who wish to utilize its Custom Neural Voice (which generates a human synthetic voice print based on a sample of someone’s speech) to explain how they will deploy it.
Microsoft has taken the first step in creating a meaningful and specific AI policy. Given the importance of AI in advancing future economic and scientific progress, Microsoft’s AI ethics policy update will likely spur its competitors to take similar steps and revise their ethics policies. Those companies that do not adopt specific language that describes how they will manage AI in a responsible manner will suffer the consequences, especially if an ethics-related issue occurs. How AI vendors’ attempt at self-regulation will play out is uncertain and could invite government scrutiny. Microsoft has shown restraint in limiting access to applications that have the potential for misuse. However, it is not certain if these policies will prevent misuse by governments. China has used AI to create a surveillance platform that monitors location, daily travel, and spending habits. Once collected, this information can be correlated to an individual’s past personal history to create predictions on future behavior. Thus, it is imperative to have an open and frank discussion about how much AI is acceptable to individuals. Once the genie is released from the bottle, it will be difficult, if not impossible, to put it back.