AI Hype: The Hidden Agenda

Beyond the dazzling promises, a new book exposes AI hype as a calculated “con” by big tech. It argues the narrative is designed to sell products, attract investment, and advance specific, often regressive, ideologies under the guise of progress.

The air crackles with promises of artificial intelligence, a technological revolution poised to reshape our world.

From the dizzying valuations of nascent startups to the breathless pronouncements of tech titans, the narrative is clear: we stand on the precipice of a new era, one defined by machines that think, learn, and perhaps, even feel.

But beneath the dazzling veneer of innovation and the sensational headlines, a different, more sober reality is emerging, one that challenges the very foundations of this grand technological vision.

Enter Emily M. Bender, a professor of linguistics at the University of Washington, and Alex Hanna, director of research at the Distributed AI Research Institute.

Their forthcoming book, “THE AI CON: How to Fight Big Tech‘s Hype and Create the Future We Want,” pulls back the curtain on what they argue is a carefully constructed illusion.

The AI hype, they contend, isn’t just spiraling out of control; it’s a calculated marketing strategy, meticulously crafted by big tech to sell products and promote specific interests.

It’s a compelling counter-narrative to the prevailing wisdom, urging us to look beyond the dazzling demos and question the very nature of what we’re being sold as “intelligence.”

At the heart of this manufactured frenzy lies the concept of Artificial General Intelligence (AGI), often presented as the inevitable evolution of today’s algorithms – a machine capable of human-level cognitive abilities, or even surpassing them.

The discourse around AGI, and its even more fantastical cousin, “superintelligence,” is deliberately vague, Bender and Hanna suggest.

It’s a nebulous term, conveniently undefined by its proponents, yet wielded with immense power to fuel investment and capture the public imagination.

OpenAI, for instance, has famously punted on defining AGI, opting to let its board decide when their algorithms have magically achieved it.

Such ambiguity, the authors argue, serves only to feed the corporate hype machine, creating a sense of imminent breakthrough that justifies ever-larger sums of venture capital.

Indeed, the claims emanating from Silicon Valley often verge on the absurd.

Remember OpenAI’s Chief Scientist Ilya Sutskever musing in 2022 that “today’s large neural networks are slightly conscious”?

Or the Google engineer Blake Lemoine, convinced his company’s language model, LaMDA, was sentient and needed legal representation?

While Lemoine was fired for leaking corporate information, not for his outlandish claims, Google’s subsequent response from VP Fellow Blaise Agüera y Arcas, suggesting LaMDA might indeed “understand” concepts, did little to quell the speculative fire.

Most recently, a Microsoft Research paper, “Sparks of Artificial General Intelligence: Early Experiments with GPT-4,” claimed the model showed the “first ‘sparks’ of artificial general intelligence,” a phrase designed to evoke the image of an imminent, self-sustaining inferno of consciousness.

Yet, this isn’t a new phenomenon.

Over six decades ago, similar bombastic predictions about computer intelligence and the specter of superhuman machines supplanting humanity were rife.

The current wave of hype is merely a digital echo of a much older tune, played louder and with more sophisticated instruments.

Perhaps the most unsettling revelation from Bender and Hanna’s work is the dark underbelly of the “general intelligence” concept itself.

Far from being a neutral scientific pursuit, the authors expose its inherently problematic origins.

The idea of identifying and measuring “general intelligence” has historical ties to racist and ableist ideologies, including the eugenics movement.

For a concept so central to the AGI dream, its definition remains remarkably elusive within the tech world.

Microsoft’s “Sparks” paper, for instance, provided a preliminary definition devoid of references to fields like psychology or cognitive neuroscience, disciplines that might actually have something meaningful to say about human intelligence.

A prior version of the paper cited a 1994 Wall Street Journal editorial, a curious source for a scientific definition, signed by psychologists who broadly defined intelligence as a “very general mental capability.”

This historical context, Bender and Hanna argue, renders the pursuit of artificial general intelligence “foolhardy at best, and deceptive and dangerous at worst.”

This isn’t just an academic debate; it’s a high-stakes financial gamble.

Billions of dollars are pouring into AGI research, much of it from venture capitalists.

While some may simply be following the latest trend, a significant contingent are “AGI true believers” – ideological billionaires like Elon Musk and Marc Andreessen, who, according to the authors, are not merely funding but actively proselytizing for a modern-day eugenics.

Their vision, cloaked in the language of technological progress, aligns with conservative politics, pro-birth policies, and a thinly veiled attack on multiculturalism and diversity.

The pursuit of AGI, in this light, becomes a vehicle for propagating a specific, often regressive, worldview.

So, why do so many involved in building these large language models seem to fall for the idea that their creations might be sentient?

And why do these same individuals often warn of “existential risk” from “superintelligence” while simultaneously pouring money into its creation?

The answer, according to Bender and Hanna, is disarmingly simple: it’s a tactic to sell AI.

In a crowded market, claims of consciousness or existential threat create buzz, differentiate products, and attract investment.

The AI ecosystem is a small, interconnected world, with researchers and founders moving seamlessly between tech giants like Microsoft, Google, and Meta, or launching startups fueled by millions in venture capital.

In 2022 alone, 24 Google researchers migrated to AI startups.

The financial stakes are astronomical; Pitchbook data reveals $41.5 billion in venture deals for generative AI firms in 2023.

McKinsey has even suggested generative AI could add “up to $4.4 trillion” annually to the global economy.

These mind-boggling estimates, while part of the hype, are powerful motivators for continued investment, regardless of the underlying reality.

This relentless hype cycle leans heavily on established tropes: sentient machines demanding rights, or Matrix-style superintelligence threatening humanity.

The implications extend far beyond the circulation of funds among VCs and investors.

Ordinary people are being told their jobs are on the line, fostering anxiety and a sense of inevitability around these technologies.

Ultimately, “THE AI CON” offers a crucial, incisive look at the technologies parading under the banner of artificial intelligence.

It dissects the drawbacks and pitfalls of this banner, revealing how the pervasive AI hype serves as a cunning cover for a small, powerful group of actors.

Their motivations are not solely about technological advancement, but about consolidating power and propagating a particular vision of the future – one that demands careful scrutiny and a healthy dose of skepticism from us all.