Greetings, it’s Yves. While I typically refrain from sharing entire posts from other authors, I feel compelled to highlight the urgent message from esteemed AI expert Gary Marcus. He has raised significant concerns about the future of AI companies, which appear increasingly unlikely to generate sufficient returns on their substantial investments. This grim reality may lead them to seek a taxpayer-funded bailout under the guise of being “too big to fail.”
Marcus has been a reliable critic of large language models, such as OpenAI’s ChatGPT, pointing out their shortcomings in fulfilling their initial promises. He has also effectively debunked the hope of achieving true artificial general intelligence. Recently, he compiled evidence suggesting that the adoption and usage of generative AI are diminishing. This trend aligns with numerous reports indicating that AI has not performed well in business contexts, despite high expectations for its applications. For example, an MIT study found that 95% of corporate trials involving generative AI failed. Similarly, research from Rand highlights that up to 80% of AI implementations fail—double the rate of other software projects—while TechRepublic estimates the failure rate to be 85%.
The notion of bailing out affluent tech executives and their venture capital backers is morally troubling, especially in light of ongoing policies from former President Trump that are exacerbating hardships for average Americans. Examples abound, such as DOGE disrupting vital programs for citizens and small businesses, including weather alerts and USDA programs that support farmers’ efficiency. Additionally, the impact of Trump-era tariffs has strained both enterprises and consumers, further complicating the economic landscape. The federal shutdown, which disrupts payments to many government workers and reduces SNAP benefits, makes any potential financial aid to the wealthy even more infuriating.
I strongly urge you to contact your Congressional representatives and share this message widely. The recent electoral successes of progressive candidates, like the Mamdani victory in New York, and the Republican losses in critical races suggest that voters are increasingly rejecting Trumpian policies. These outcomes should alarm both Republican incumbents and any Democrats siding with Trump, making it essential for constituents to express their disapproval of an AI bailout as a perilous political choice.
By Gary Marcus, professor emeritus of psychology and neural science at New York University. Originally published at his website
Recently, Sam Altman expressed frustration when asked by Brad Gerstner about OpenAI’s ability to meet $1.4 trillion in obligations with only $13 billion in revenue.
In his lengthy but largely unconvincing response, Altman claimed OpenAI was generating more revenue than reported, criticized the questioner, and vowed that future earnings would be impressive.
“First of all, we’re doing well with more revenue than that. Secondly, Brad, if you want to sell your shares, I’ll find you a buyer. I believe there are many interested parties who would love to purchase OpenAI shares. People concerned about our computing operations would be thrilled to buy. We plan for revenue to increase significantly. Revenue is growing rapidly, and we are betting that not only will ChatGPT continue to expand, but we will establish ourselves as one of the leading AI clouds and that our consumer device business will also flourish. Our technology will automate scientific processes and create immense value. We plan carefully and understand where technology is headed, as well as the products we can develop and the revenue they will generate. Although we might make mistakes, we are committing to this risk.”
What Altman did not disclose at the time was his intention to decrease borrowing costs by potentially having American taxpayers cover the expenses.
Recently, at a Wall Street Journal conference, OpenAI’s CFO hinted at the possibility of significant indirect government support for AI companies:
In justifying what could be one of the largest indirect government subsidies in history, the CFO remarked, “AI is nearly a national strategic asset. We must approach our competition with countries like China deliberately.” (Nvidia seems to be planning a similar move.)1
Recall this tweet?
Almost ten months ago, we witnessed this very scenario unfold, and rumors suggest that the government might agree to this approach.
This means that taxpayers will bear the burden.
This situation is unacceptable. Today, I urge you to contact your Congressperson and express your opposition to using taxpayer dollars to support overhyped and financially unstable AI companies that consistently overspend. Workers who are already suffering from layoffs should not be the ones footing this bill.
Let’s act proactively to prevent this ‘too-big-to-fail’ narrative from becoming irreversible.
