Artificial Intelligence: Leveraging Benefits and Anticipating Unintended Consequences

Artificial intelligence (AI) has quickly become a regular topic of conversation and viewed as a magical genie by some. It is in the news on a daily basis and being considered by organizations who want to leverage this powerful tool for greater efficiencies and streamlining processes. The fact that this technology is so new, with so many purveyors of different products with little legal or regulatory oversight makes it ripe for unintended consequences. It is critical that businesses consider the ramifications of implementing (or not implementing) AI within their organization.

As ubiquitous as AI has become in our daily lives, it still warrants a discussion of what AI entails. Essentially, it is computing based upon machine learning in order to make decisions in a human-like way. It can efficiently analyze data by identifying patterns, and utilizes those patters to offer predictions. It can be leveraged for a variety of different uses, including computer network threat monitoring, healthcare diagnostics, writing computer code and drafting speeches or business documents. AI holds a tremendous amount of promise but it also can lead to unintended consequences when not thoughtfully implemented.

There are a multitude of examples within the last year alone where individuals and organizations were negatively impacted by the reckless implementation of AI. Although some of the loopholes for abuse or improper application have been responded to by at least the larger AI product providers, there may be other consequences not yet seen.

By way of example of the significant inaccuracies of AI, a law professor was wrongly accused of sexual misconduct. ChatGPT wrongly included him in a list of professors who allegedly committed sexual misconduct against students. Not only was the professor’s inclusion on the list entirely false, the ChatGPT-generated narrative included citations to fake articles, including one allegedly from the Washington Post. The article did not exist, nor did the scenario in which the alleged incident took place.

There have also been several examples of lawyers improperly utilizing AI to draft motions that were filed with the court containing fake case law citations. Only after the courts attempted to review the cited case law was it discovered that the cases and the premise they purported to support were fabricated. This had dire consequences for those lawyers who faced suspension of their law license or disbarment.

There are also examples of employees misusing AI in an effort to make their work lighter and create a better work product or solve a vexing problem. One company had its sensitive information disclosed when a well-meaning employee attempted to use AI in order to draft board minutes. In another example, a Samsung employee entered proprietary code into AI to fix a bug and, thus, rendered it publicly available.

There are further consequences with respect to intellectual property. OpenAI, for example, allegedly utilized more than 300 billion works derived from all manner of media from websites, published articles, books and blogs to enhance its product through machine learning. However, OpenAI failed to obtain permissions for this use and there is an ongoing dispute over whether this creates a copyright issue or whether the materials were subject to fair use.

Similarly, there are concerns over the creative process and utilization of AI. Recently, there were two concurrent strikes in the entertainment world involving writers and actors. The disputes were, in part, due to the concerns over entertainment companies using AI in ways that would impact these creative workers by creating AI likenesses of actors from prior data or using AI to write scripts — both of which would obviously have detrimental effects for those actors and writers.

There are also positives. AI can protect organizations from network attacks by monitoring systems and identifying anomalies and certain behaviors of malicious actors. This can be more comprehensive and faster than simply relying on known code or upon human review.

AI can also be used to speed up diagnosis in the healthcare space. It assists medical providers in analyzing patients’ images, for example, to more effectively identify indicators of breast cancer or other conditions. AI can also be used proactively by analyzing vast swaths of data from genetic testing, medications, diagnoses and images to identify correlations that may better and more accurately identify individuals who are at high risk for a certain health condition.

The manufacturing sector could also benefit from utilizing AI. AI can continuously monitor the production floor and its machinery to instantly identify issues before they become catastrophic. For example, AI can identify heat spikes or vibrations in machinery and alert to the need for maintenance before the machine breaks down, causing more damage and delay in production.

AI is a tool with great potential for businesses to streamline their processes, mitigate downtime, more effectively assist patients and customers and more. However, implementation requires great care and thought. Data privacy concerns and policies and procedures must be articulated and addressed before the genie is let out of the lamp.

JD, MS CyS, FIP, CIPP/US, CIPP/E, CIPM | + posts

Blair Dawson, JD, MS CyS, FIP, CIPP/US, CIPP/E, CIPM, brings extensive experience in data privacy consulting and responding to data security incidents to McDonald Hopkins' Data Privacy and Cybersecurity Practice Group.