Debunking the Myths of Artificial Intelligence: Insights from Experts
Linguist Emily M. Bender from the University of Washington and engineer Alex Hanna, formerly of Google and currently the director of research at the DAIR Institute, are challenging the inflated narratives surrounding artificial intelligence (AI). Their new book, The AI Scam, released by Paidós in Spain, aims to demystify AI, which they prefer to refer to as “synthetic content generating machines.”
A Critical Look at AI's True Nature
The authors assert that AI is not genuinely intelligent and claim it is often misrepresented as a solution to numerous problems. “We will be honest with you: AI is a trap, a scam, a deception being sold to enrich certain individuals,” they declare, setting the tone for a comprehensive critique that spans over 300 pages.
Bender, who also directs a master's program in Computational Linguistics, coined the term “stochastic parrot” to describe large language models (LLMs), illustrating how these systems probabilistically regenerate the data they were trained on. Hanna, who resigned from Google in protest after the firing of AI ethics leader Timnit Gebru, emphasizes the financial and ethical ramifications of AI technology.
The Investment Bubble in AI Technology
Hanna highlights a concerning trend: significant investments are being funneled into AI technology, with projections estimating that by 2025, $50 billion—nearly 70% of all investments in North America—will target AI. “The advertising has been constant,” she states, pointing out the dangers of a narrative that positions AI as indispensable.
Bender adds that this perception drives investors to overlook potential risks. “Why invest in a company unless it's believed to be on the verge of creating a machine that can literally do anything?”
Shifting Perspectives on AI
The authors argue that narratives from both AI advocates and skeptics often converge, further complicating public discourse. “Those enamored with AI believe it can resolve all our challenges, while the skeptics fear it will lead to our demise,” Hanna notes. They contend that both viewpoints ultimately accept the inevitability of AI's rise, albeit with opposing conclusions.
Racist Foundations of General AI Aspirations
Bender and Hanna raise ethical concerns regarding the pursuit of General AI (AGI), asserting that its origins are often entangled with eugenics and biased intelligence testing. “The framework driving the search for AGI includes figures who were proponents of intelligence classification based on race,” Hanna points out, calling for deeper scrutiny of the underlying motives behind AI advancements.
Addressing Current AI Issues
The authors stress that focusing on futuristic doomsday scenarios detracts from immediate concerns related to AI's impact on society. “Geoffrey Hinton, who developed pivotal AI algorithms, emphasizes global risks, yet the few affected individuals are often economically disadvantaged or from marginalized communities,” Bender argues.
Advocating for Responsible AI Development
To combat the growing AI bubble, the authors suggest resistance and awareness. Bender advocates for a concerted effort against the narrative of AI inevitability. “We can resist this notion; the future is not predetermined,” she asserts. They also believe that Silicon Valley workers, if united, could influence the course of AI development significantly.
