Is Artificial Intelligence Value-Neutral?

Over the past summer, I created a module for Geneseo’s Ideas that Matter program on AI. In it, I set out to give a starting point for anyone interested in the broad topic of whether technology, specifically Artificial Intelligence, is “value-neutral” or “value-laden.” I began thinking about the question of value-neutrality while teaching a class at Monroe Community College called “Technology and Values.” (I’m currently teaching a version of it as an Honors course and I’ll be teaching it as a Philosophy course in Spring 2025.) Often, when I would ask my students to reflect on whether technology has values embedded in it, they would respond that technology is neutral; it is simply a tool—neither good nor bad. It is the user who defines the “value” of the tool and so it is the user who is to blame for technology’s failings.

I can sympathize with this commonsense reading. Many of our pieces of technology are just tools that don’t seem to bias any particular value. Is a chair “good” or “bad”? A paint brush, a car, a smartphone, a social media platform? What about a knife, an AR-15, a nerve agent or a nuclear bomb?

I don’t deny the plausibility of the above position, but I think it is most tenable if we understand technology to be merely objects, machinery, or products. If technology is a system or systems, consisting of objects, machinery, and products (yes), but also types of background knowledge, rules and norms, users and creators, we can begin to see how bracketing value from technology is a hopelessly misguided attempt. Best case, the notion might simply deflect the issue; at worst, it leaves us (consumers, users, victims, the general populace) to bear the costs, deceiving us into thinking it is the user who is wholly to blame, and making us believe that the only answer can be being better humans and improving the ends to which we put the objects, machinery, and products. This is exactly the sort of scenario that occurred with social media, and I see a similar pattern playing out with AI.

I’m still sussing out my view so I’d love to hear feedback and counterarguments.

Below are two excerpts from my Ideas that Matter module. The entire module, “AI and Value-Neutrality,” can be found at Knight Scholar.

Artificial Intelligence is on the tips of everyone’s tongues these days – What exactly is it? What can it be used for? What will it be used for in the future? What problems will it create or solve or exacerbate? This learning module aims to look at a specific facet of AI — the issue of value-neutrality — by having students look inward at capabilities necessary for human flourishing and then ask whether AI can cultivate (or inhibit) those capabilities. This will lead to a discussion of what values underlie AI and what this says about whether or not AI is value-neutral. To help excavate all these questions, students will read scholarly and non-scholarly texts, watch informational films, and reflect on a case study involving bias and AI.

 

The question central to this module is whether Artificial Intelligence is “value-neutral”, but this question is a bit of a red herring as it leads to more and more questions such as: What does it mean for a technology to have “value”? And what’s more, what does it mean for anything to have “value”? Is it the same as being valuable? What is technology? Can we answer the question of value-neutrality for all technologies or does an answer have to be particular and context specific? And on and on. This is, no doubt, the way philosophers think. Which I am, and that is what guided the flavor of this module. My aim with these notes is to present an instructor with the borders of the debate around value neutrality so they can focus on different areas if they so choose.

Image credit: “Something of Value [34 of 52]” by CJ Isherwood is licensed under CC BY-SA 2.0 .