[The AI Hype Wall of Shame is a collaboration between Critical AI and the DAIR Institute. For more on this initiative and our rating system, see this link.]
Just as the U.S. has an annual State of the Union, so the European Union has a State of the European Union (SOTEU) address. Every September, the President of the European Commission addresses the European Parliament. They usually discuss the economy, political vibes, some proud moments from the past year, and of course, what’s next on the agenda. This year, however, President Ursula von der Leyen devoted nearly five minutes to artificial intelligence (here, 32:15-37:00), during which she included multiple, predictable AI end-of-the-world narratives. Welcome to the club, Madam President. You’ve just earned your spot on the #AIHypeWallofShame.
As Google research scientist Francois Chollet remarked on social media, “the extinction risk from AI, much like extinction risk from self-replicating grey goo nanobots or from an alien invasion…should be pretty far down the list of global priorities.” In the midst of actually existing crises in Europe, President von der Leyen found time dive headfirst into the AIHype pool. We give her a rating of 3 Marvins—our very highest!—for a full-blown leap into mind-blowing paranoia.
Von der Leyen began with the tried-and-true “AI is gonna revolutionize everything” mantra. A drum roll for everything AI will miraculously fix: healthcare, productivity, and, of course, climate change!
Let’s pause to review some earlier claims about AI’s miraculous powers to fix healthcare. In 2016 Geoffrey Hinton issued his notorious “AI will replace radiologists” prophecy, possibly precipitating a shortage in a field that now has more unmet need for human expertise than ever. But, hey, it’s just a roll of the dice on whether ailing people need potentially life-saving surgery or not. I’m sure an all-purpose genius like Hinton can weigh in on the matter with no need for an M.D., residency, radiology certification, or any other pesky criteria for knowing what one’s talk about. So why worry about over-hyping AI?
This should make us wonder who von der Leyen is talking about when she declares that “We need [AI],” who is the “we” in question? When healthcare professionals are left to clean up after the bad guesses of self-promoting tech bros, we can be sure that “we” certainly doesn’t include the artists and writers whose copyrights and creativity have been mowed over by tech giants keen to usher in a new for-profit paradise for “AI Art.” Please tell us Madam President, who are you speaking for?
But this was just the warm-up!
Having fawned over AI’s unproved marvels with sufficient gusto, it was time to go full on doomer.
“We also should not underestimate the very real threats coming from AI,” von der Leyen intoned, dutifully quoting the Center for AI Safety’s “warning” that: “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.” (here, 32:35-33:00). Uh, hold up a sec.
Here von der Leyen pays homage to the rallying cry of a motley crew of Singularitarians, Doomers, Longtermists and a few others who may have blundered into the party unknowingly. Though the goal is clearly to elevate AI by whatever means possible, left unclear is how AI could be as threatening as pandemics or nuclear war–actually existing threats with a documented history of causing harm to the peace and prosperity of billions of people. Maybe von der Leyen has some evidence that she chose not to share: for now we’ll vote with Chollet and urge the EU to weigh in next on the mortal threat of alien space invasion.
And what happened it to climate change? The answer of course is that AI boosters don’t like to foreground a problem that might lead the public to inquire into the vast energy and environmental resources that this technology requires.
Much of this “AI is going to end us all” narrative is a cocktail of wild overestimations mixed up by tech leaders with more money, power, and lackeys than either education or common sense. They get their ideas from sci-fi plots, magical thinking, and some downright confusion about humans, machines, and the known facts of science.
It’s disturbing to see a leader of von der Leyen’s stature (perhaps joined by the UK’s Rishi Sunak), regurgitating such rhetoric outright. Could von der Leyen give a concrete example of the type of AI-led catastrophe she’s discussing in the same breath as pandemics and nuclear war? Could she wheel out a scientific expert or two? By embracing this narrative, von der Leyen enables tech companies to manipulate the kind of regulation the public needs.
In the real world, “AI” is exacerbating vast power imbalances, growing inequalities, and entrenching the exploitation of workers. Add to that, the immense environmental footprint of some AI practices, the violation of data privacy, and copyright infringement, and we have a situation that cries out for serious oversight. It’s frustrating, to say the least, when these concerns are drowned out by science fiction; experts on actual harms don’t even get a seat at the table.
In the midst of the turbulence von der Leyen’s remarks stirred up, there is a glimmer of hope: the AI Act proposal does not seem to mirror the views in her speech. This is a testament to the diligent work of those who have been dedicated to crafting legislation that is both practical and grounded in reality.
While a commendable step forward, the European proposal is not without shortcomings. There’s an undeniable need for more comprehensive protections for the rights of those impacted by data-driven “AI” systems. Clearer boundaries must be established around unacceptable applications, including its potentially harmful use in areas like migration.
The rise of generative AI also necessitates more stringent regulations, ensuring responsible use and minimizing the flagrant ethical problems that plague this technology.
Nevertheless, the current AI Act proposal stands as a beacon against the kind of irresponsible hype that von der Leyen chose to voice. It underscores the importance of a legislative framework grounded in tangible realities rather than dystopian fearmongering. The key will be to maintain this trajectory, ensuring that the legislation remains unswayed by sensationalist demagoguery.
Let’s keep the EU off the wall of shame and on the terrain of scientific fact and democratic oversight in the public interest.