OpenAI Says “AGI” Is Here — What Does That Actually Mean?

SUMMARY

Ed Elson hosts a discussion with Gary Marcus on the validity of OpenAI’s AGI announcement and Kathryn Anne Edwards on the implications of the August U.S. jobs report. The episode concludes with analysis of Germany’s recent election and the rise of the far-right AfD party amid economic stagnation.

MAIN POINTS

• OpenAI releases Astra, its largest and most expensive model, prompting industry leaders to declare the arrival of AGI. [1:01]
• Gary Marcus critiques the shifting definitions of AGI and argues that Astra does not meet historical or economic expectations for artificial general intelligence. [5:01]
• Discussion centers on the use of AGI as a marketing term, highlighting the lack of real-world economic impact and the tendency for such announcements to influence stock prices. [10:01]
• The August U.S. jobs report shows stronger-than-expected job growth, but underlying labor market indicators remain weak. [19:00]
• Kathryn Anne Edwards identifies persistent issues in the labor market, including long-term unemployment, weak wage growth, and declining labor force participation among prime-age men. [26:00]
• Edwards rebuts claims that U.S. workers take too many holidays, noting that the U.S. lags behind peer countries in statutory paid leave and worker protections. [30:01]
• Germany’s far-right AfD party achieves a historic electoral victory, attributed to economic stagnation and public frustration with the status quo. [35:01]
• Analysis concludes that Germany’s economic woes stem from structural issues, not immigration, and warns that scapegoating immigrants is a misguided and counterproductive political strategy. [37:01]

DETAILED ANALYSIS

OpenAI’s release of the Astra model has reignited debate over the definition and reality of artificial general intelligence (AGI). Astra, trained on over 100,000 Nvidia GPUs at significant expense, is celebrated by industry figures such as Nvidia CEO Jensen Huang and OpenAI President Greg Brockman as a generational leap, with some declaring that AGI has arrived. However, this claim is met with skepticism by experts and observers who note that the anticipated economic and social transformations associated with AGI—such as plummeting wages, mass unemployment, and the automation of complex tasks—have not materialized.

Gary Marcus, an author and emeritus professor at NYU, provides a critical perspective on these developments. He highlights the phenomenon of 'benchmark maxing,' where AI models are optimized to perform well on specific tests but often fail to deliver consistent results in real-world applications. Marcus points out that each new model, including Astra, is hyped as a breakthrough, yet the practical impact remains limited.

He refers to the 'CLA effect,' where companies announce plans to replace human workers with AI, only to quietly rehire them when the technology falls short. This recurring pattern suggests that the technology, while advancing, does not yet fulfill the broad capabilities implied by the original definition of AGI.

The discussion delves into the shifting goalposts for what constitutes AGI. Historically, AGI was defined as a system capable of performing any cognitive task a human can do. Over time, this definition has been diluted, with some now suggesting that performing most economically valuable work suffices.

Marcus argues that current models do not meet even these weaker standards, as they struggle with tasks that require reliability and adaptability—qualities intrinsic to human intelligence. He also critiques the use of AGI as a marketing term, noting that repeated declarations of its arrival serve more to boost company valuations and stock prices than to reflect substantive technological progress. The lack of clear, consistent criteria for AGI allows for ambiguous claims that ultimately erode the term’s meaning.

The conversation transitions to the August U.S. jobs report, which shows the addition of 162,000 jobs—far exceeding economists’ expectations. Despite this headline figure, labor economist Kathryn Anne Edwards notes that underlying indicators reveal persistent weaknesses. The unemployment rate remains steady at 4.1%, but wage growth is lagging behind inflation, and the number of long-term unemployed is elevated.

Edwards emphasizes that month-to-month fluctuations in jobs data are often overinterpreted and that the broader trend points to a labor market that is weak and gradually slowing. She highlights the decline in labor force participation among prime-age men and the lack of robust wage growth as signs that the labor market is not delivering widespread benefits.

Edwards also addresses recent political commentary criticizing the number of holidays in the U.S., arguing that the country actually offers fewer statutory paid holidays and vacation days than its peers. She attributes this to declining unionization and weakened worker protections, which have contributed to stagnant wages and diminished labor power. Edwards contends that the real issue is not excessive time off but the failure of policymakers to implement measures that would improve workers’ economic security.

The episode concludes with an analysis of Germany’s recent regional election, where the far-right Alternative für Deutschland (AfD) party achieved its best result since its founding. The AfD’s rise is linked to prolonged economic stagnation, declining manufacturing jobs, and falling investment. The party’s platform centers on aggressive anti-immigration policies, which are presented as solutions to Germany’s economic challenges.

However, the analysis points out that the true causes of Germany’s struggles are structural: conservative fiscal policies, regulatory burdens, energy dependence, and demographic decline. Blaming immigrants, the analysis argues, is a politically expedient but fundamentally flawed response that echoes the dynamics seen in the UK’s Brexit referendum. The warning is clear: scapegoating vulnerable groups does not address underlying economic issues and risks further harm to national prosperity.

LINKS

• Prof G Markets YouTube channel
• Prof G Markets newsletter subscription
• Order 'Notes On Being A Man'
• Subscribe to No Mercy / No Malice
• Prof G Markets on Instagram
• Scott Galloway on Instagram
• Ed Elson on Instagram
• Ed Elson on X (Twitter)
• Ed Elson on Substack
• Vanta compliance platform
• Delete Me data removal service
• Indeed sponsored jobs

KEYWORDS