By Nathan Elliott, Director of Professional Services, J2 Interactive
LinkedIn: C. Nathan Elliott
LinkedIn: J2 Interactive
Data centers are the talk of the town these days. Are they an exciting advance in delivering better care, faster? Are they an environmental disaster? Are they a pawn in global political conflict? Or, are they something else?
The polarization over AI is real and important, and it should be acknowledged. It seems you’re either on the leading edge and celebrated for doing new things, or you’re the villain for numerous reasons, including the environment, neighborhood pollution, and feeding the behemoths that talk of potentially dubious uses of technology for all of society. To successfully navigate this, we need to find a path through the emotions that enables the enterprise to advance, while making sure the road we choose is as sustainable and emotionally neutral as possible.
In March 2024, the Securities and Exchange Commission (SEC) issued its final Climate-Related Disclosure Rule, which mandates that certain public companies must begin disclosing specific environmental costs in their financial statements and annual reports. Our society, rooted in capitalism, has often been accused of sweeping environmental costs under the rug (“externalizing” or “socializing” them) by not including these costs in financial reporting. We focus on dollars in and dollars out, but not species killed, waterways damaged, or trees cut down. Unfortunately, the SEC rule is effectively paused under the current administration. However, if the pendulum swings the other way after the upcoming midterm elections, this rule, and others like it, could become much more impactful over the next few years.
In that scenario, organizations would have to think about a possible shift toward broadening accounting definitions to include these “externalities” – the ripple effects of doing business that are not directly captured in accounting records or currently reported to the public. Accounting bodies in the U.S. frequently make updates to accounting rules and definitions, and doing so is nothing new. What would be new is requiring definition changes that start to more accurately capture costs that society has ignored for a very long time.
Whether by formal policies in the years to come or by social pressure in the here and now, AI usage is most assuredly under well-deserved scrutiny from all sides.
For health systems, health information exchanges (HIEs), and other technology companies serving healthcare professionals and their patients, we have the added societal expectation of “do no harm.” So, how do we reconcile all of this and find a way to still leverage new technology like AI and tap into its benefits while remaining sensitive to the potential harm to which we could be contributing?
Here are a few things to think about:
- AI without Data Centers. Not all AI requires the use of highly controversial infrastructure, like “data centers.” In fact, AI can be architected to run on local computers and tap into central data stores through standard connections. Organizations should evaluate how they are currently set up to use AI and how it is delivered. Have you explored lower-impact alternatives?
- Some Uses of AI Are Better than Others. Think about the ways you plan to use AI. Conducting text-based searches or summarizing content held in documents or databases is far less energy-intensive than using AI to generate videos or music. If you establish an internal policy that minimizes the energy demands of your employees’ AI use, you can highlight this in your communications and call attention to how your usage compares to other things. For example, you can show how your organization’s energy use for AI compares to something conceptually tolerable such as the energy consumption of an average American home over a certain period of time, or on a smaller scale, that it equates to running a microwave oven for a certain number of minutes per day.
- Explore Where Your Electricity Comes from. If you use the first example above and intentionally avoid using data centers for your AI consumption, you might also be able to tell the story of how the energy that your AI solutions do use comes from sustainable sources (if it does).
- Explain the Gains. If you are able to do so, think about measuring and tracking the benefits from using AI to compare them with other ways of generating similar results. For example, you might quantify the people-hours required to complete certain tasks, or you might quantify the increased revenue or cost savings associated with leveraging AI for tasks such as data quality improvements or reducing billing errors. It is important, though, to continue to report these benefits compared to costs incurred. Showcasing achieved benefits alone will be perceived as “rosy” or one-sided. If you can prove that AI helped save a life, even better. The industry needs more positive stories.
- Control Its Use. You may also want to consider only allowing AI to be used in certain ways for certain specific tasks within your organization. Your AI policy might consider patient care-related tasks to support providers as an acceptable use of AI, whereas, running security audits or conducting financial analyses is not worth the potential energy consumption or assumed environmental impact. Tailoring AI use to only break eggs when you really need an omelet should reasonably be somewhat more tolerable to AI critics.
As you think about AI use and governance policies in your organization, it is helpful to remember that your stakeholders likely hold very strong views of AI and data centers right now. Fortunately, you have some options to proactively steer your AI use and messaging about it in ways into less controversial territory.