Let’s call AI something else
“AI” doesn’t mean anything. As Emily M. Bender and Alex Hanna write in their new book The AI Con:
To put it bluntly, “AI” is a marketing term. It doesn’t refer to a coherent set of technologies.
The significance of the term resides not in what AI is, but what the word is intended to do: that is, to get people excited about “AI.” In a weird sort of circular logic, the term is meant to generate interest in the term. To say it another way:
Say the magic word, so that people get excited about the magic word.
So many people are invested – literally – in this idea. Venture capitalists, startups seeking funding, established companies trying to recover some relevance, and above all, the unethical Big Tech monopolies, continuing to pursue their solitary goal of growth at any cost. All of them are cramming AI into every product, service, feature, menu, toolbar, app, platform, and workflow, whether users want it or not, all so that they, too, can say the magic word.
The boosters present AI as an all-purpose solution, a balm for every social ill. We hear, for example, that “AI will solve climate change,” even as AI data centers are measurably accelerating climate change.
This calls for a takedown. Bender and Hanna have provided one in The AI Con:
Every time we write “AI”, imagine we have a set of scare quotes around it. Or if you prefer, replace it with a ridiculous phrase. Some of our favorites include “mathy maths”, “a racist pile of linear algebra”, “stochastic parrots” (referring to large language models specifically), or Systemic Approaches to Learning Algorithms and Machine Inferences (aka SALAMI).
Try it out. “SALAMI is going to change everything!” Or as Stefano Quintarelli put it back in 2019:
Will SALAMI develop some form of consciousness?
Will SALAMI have emotions?
Can you possibly fall in love with a SALAMI?
These are all questions that have been asked about AI. Just how ridiculous they are is so much clearer when asked, instead, about SALAMI.
Apart from the “ridiculous phrases,” Bender and Hanna argue that much of what people understand as AI can be described as automation. And as the 19th-century Luddites pointed out, automation isn’t necessarily bad. It’s the systems that exploit workers, and deliver shoddy products to customers, that need to be opposed. Unfortunately, that describes much of AI today.
The reality, of course, is that digital automation can be positively helpful. Bender and Hanna give the example of the automated blood-pressure monitor in doctor’s offices: by handling a rote task that doesn’t require human judgment, the machine frees up the medical staff for more important work. Our world would be so much better if digital automation was built for helpful outcomes, rather than the exploitation of the many for the enrichment of a few.
I spoke with Bender and Hanna this week on Techtonic about The AI Con:
Listen to the show (interview starts at 3:00)
Episode page with links
Bender (a co-author of the “stochastic parrots” paper from a few years ago) and Hanna are smart and engaging, describing why we need to look beyond AI hype in order to chart a better path with technology. Recommended.
As I mentioned at the end of the interview, one of my favorite lines in the book could have been the subtitle:
We can’t technology our way out of it.
Whatever positive change we want to see in society – more opportunity, more justice, less exploitation – we’re not going to get there with an overhyped basket of platforms called “AI.” Technology can be helpful, essential even, to our plans – but as a tool, not the focus of our efforts. The empty promises about AI we’re being sold are nothing but a con.
This week on the Forum:
If you join Creative Good you’ll get access to our members-only site, the Creative Good Forum, which has these recent posts I’d recommend:
Summaries of new critiques of AI: Karen Hao writes about the need for “journalists, civil society, researchers, policymakers, and citizens to push back against the tech giants” – while the AI Now Institute has issued a new 2025 Landscape Report.
Updates about scammers being enabled by Big Tech: Facebook/Meta profits from scammers’ activity but takes no responsibility and accepts no liability; and Elon Musk’s Starlink “enables a multibillion-dollar online scam industry,” writes the Sydney Morning Herald.
Speaking of Starlink, Musk’s satellites are re-entering the atmosphere earlier than expected due to increased solar activity. Link
Email: alternatives to Gmail is a new post in Good Reports, my resource for finding alternatives to Big Tech platforms. (I recommend Fastmail and several other alternatives. Join Creative Good to get access.)
Until next time,
-mark
Mark Hurst, founder, Creative Good
Email: mark@creativegood.com
Podcast/radio show: techtonic.fm
Follow me on Bluesky or Mastodon