How to move away from the untenable, uncomfortable middle ground

By Emily M. Bender and Decca Muldowney

Amid all the interminable discussions of “AI” these days, we’ve noticed a new pattern emerging: people trying to occupy an untenable middle ground with regard to the use of systems sold as "AI". This is a position where people try to recognize the harms of this tech but also hold space for "responsible" or "ethical" use.

We’ve also noticed that when someone is trying to hold this untenable position in a discussion, a few things tend to come up (not every time, not everyone):

1. Defensiveness. People read criticism of the systems and proposed uses of the systems as accusations that users are "bad people". Thus a criticism of the tech lands as criticism of the user, and tensions flare. Our critique of “AI” is not based on the personal ethics of the user, but the problems embedded in the systems at large, the companies behind them and governments more and more in the thrall of those companies.
2. Righteousness. People do have legitimate needs, often unmet needs, and the synthetic text extruding machines can look like a solution. A common argument is that “AI” will benefit vulnerable or disenfranchised users in various ways, such as making health advice more accessible or “democratizing” education, or any of the variety of “solutions” to systemic accessibility problems that Rua Williams dubbed “disability dongles”. But just because the problems are real doesn't mean the solution is beneficial, effective, or worth its (not always externalized) costs. Unfortunately, pointing any of this out is taken as the same as saying you don't care about the legitimate needs.
3. Whataboutism. This is an accusation of hypocrisy that is used to brush off concerns about the externalities of these systems. You eat meat, you fly on airplanes, etc, etc, so how dare you talk about the impacts of data centers? This is a recipe for inaction, on any front.
4. Tone policing. People who are trying to occupy that uncomfortable, untenable space will claim that clear statements of harms or strong principles against use of these systems will "turn others away" as if the centrists are the ones actually pushing for more ethical practice. But this "other people won't listen" remark strikes us as a way of saying: "This makes me uncomfortable", while trying to claim to be on the right side of history at the same time.
5. Wishcasting. Some folks will point to scientific results from fields outside their own that are marketed as having been done with "AI" and ask: How could you take a hard line against "AI" when it has provided XYZ? Such remarks both conflate many different things under the umbrella of “AI” and are also usually based on hype-filled media coverage that tends to obscure what technology was actually used in what way.
6. Exceptionalism. "I know this can be dangerous for people in general, but I know how to use it carefully."/"I know how to verify every output, and I am not deskilling myself." But in that case, how could such users know they are safe? This position also mixes acknowledging the danger to others while setting a risky example in talking up one’s own use.

So what is the best way out of that uncomfortable, untenable space? We think one key step is disaggregating the (non-coherent) set of technologies sold as "AI". If you don't fall for the marketing gimmick telling you all this stuff you work with is "AI", you aren't saddled with trying to defend any of the rest of it.

A good example of this is a recent conversation Emily had that turned in part on a strange, over-expansive definition of "genAI" which included, for example, optical character recognition (OCR).

OCR can be a useful tool for many research projects! (Decca notes it’s incredibly valuable for journalists and has been used by data reporters for years, long before our current “AI” hype cycle.) OCR is also the kind of technology that gets better with better language models, i.e. more fine-grained models of which word(parts) go where. That has been true since before "genAI" and will be true after.

Just because you can use the synthetic media extruding machines to approximate the task of OCR, however, doesn't mean that that task can or should be used to justify the use of "genAI" in research or journalism.

Another important step is a values examination. What is important to you? How are those values supported or not by entering the discourse in a way that holds space for OpenAI/Anthropic/Google/Meta and all the others in this massive push to shove "AI" into every part of our lives as "not all bad"?

For scholars we suggest asking: What are your research goals, what do you value about participating in scholarship, how can you meet those goals/act in accordance with those values and what obstacles are in your way? For journalists: What are you hoping to achieve with tools sold as “AI” that cannot be done with traditional reporting methods or data journalism techniques that already exist? Is that outcome so valuable that it outweighs the ethical problems of using these tools? What is your evidence that the tools function as advertised? More broadly, for anyone who finds themself reaching for a chatbot or other technology associated with well-documented harms, what is the need that you are filling with that use? What values come into play, and how else might that need be met, or at least partially met, in accordance with those values?

Part of what makes that middle ground untenable and uncomfortable is that it requires carrying water for these clearly bad actors. The good news is you can set that bucket down and step out onto firmer ground.

This does require going against the mainstream, but that gets easier when a) you find you're not alone and b) you see how much of mainstream opinion on this is actually the result of marketing.

In the Mystery AI Hype Theater 3000 podcast, we are frequently exposing AI hype for the marketing that it is, even when it is done by journalists and others on behalf of the companies pushing their products. For specific discussions of “AI” and science, journalism, and software engineering, check out these episodes of the podcast:

Episode 31 - Science Is a Human Endeavor: Will AI someday do all our scientific research for us? Not likely. Drs. Molly Crockett and Lisa Messeri join for a takedown of the hype of "self-driving labs" and why such misrepresentations also harm the humans who are vital to scientific research. [Livestream, Podcast, Transcript]

Episode 65 - Crunching the Numbers: So-called AI tools are increasingly infiltrating newsrooms, particularly when it comes to data analysis. DAIR writer-in-residence Decca Muldowney joins us to discuss the need for journalists to distinguish between "AI" and reliable, verifiable research methods. [Livestream, Podcast, Transcript]

Episode 79 - Beware the 20x Engineer: "Sure, LLMs are bad at some things, but you can't deny that they're useful for programming!" Sound familiar? In this week's episode, Emily and Alex break down the key myths around AI-boosted productivity in tech. [Livestream, Podcast, Transcript]