OpenAI Warns of Post-AGI World, Proposes Economic Overhaul

OpenAI Issues Stark Warning About Post-AGI Future

OpenAI has released a paper outlining a future where Artificial General Intelligence (AGI), systems that can perform at or above human cognitive abilities, could lead to massive job losses and societal disruption. The paper, titled “Industrial Policy for the Intelligence Age,” suggests that current economic and safety structures are insufficient for the coming era. OpenAI believes superintelligence is no longer a distant concept but a near-term reality, prompting them to propose solutions proactively rather than face regulation after potential damage occurs.

The Accelerating Pace of AI Development

The rapid advancement of AI is a key concern. OpenAI aims to launch research interns by 2026. These AI systems will be capable of reading scientific papers, identifying research gaps, and suggesting next steps.

By March 2028, the goal is a fully autonomous AI researcher that can design, conduct, and present experiments without human intervention. This accelerated research cycle could dramatically speed up AI development even further.

Sam Altman, CEO of OpenAI, has stated that the company is “close enough to AGI that the precise definition matters.” Some within OpenAI believe they may have already reached this milestone. Other AI leaders share similar timelines.

Anthropic CEO Dario Amodei predicts AGI could arrive by 2027, while Demis Hassabis of Google DeepMind gives it a 50/50 chance by the end of the decade. These are not fringe predictions but come from those deeply involved in AI’s cutting edge.

Economic Disruption Looms

The potential impact on employment is significant. Goldman Sachs estimates that generative AI could affect 300 million full-time jobs globally. McKinsey research indicates that 57% of current U.S. Work involves tasks that today’s technology could automate.

While the World Economic Forum projects more new roles created than displaced by 2030, the transition will be difficult. The people losing jobs and those filling new ones will likely not have the same skills or be in the same locations.

Data already shows early signs of this disruption. Employment for 22 to 25-year-olds in AI-exposed roles has fallen, with a nearly 20% decline among young software developers.

Entry-level job postings have dropped significantly since early 2023. These figures are based on current AI models, and the impact is expected to steepen dramatically with the arrival of more advanced systems like autonomous AI researchers.

OpenAI’s Proposed Solutions

In response to these challenges, OpenAI’s paper proposes several policy ideas to manage the transition and ensure people remain central to the economy.

1. Public Wealth Fund

OpenAI suggests the U.S. Government create a national investment fund. This fund would be financed by AI companies themselves. Every American citizen would receive a direct financial stake in AI-driven economic growth, similar to Alaska’s Permanent Fund which distributes oil revenues to residents.

The fund would invest in AI companies and businesses adopting AI, with returns distributed to citizens. This is seen as a form of Universal Basic Income (UBI) funded by the very companies driving automation.

2. Robot Taxes

The paper addresses the potential crumbling of the tax base as AI automates work. Since many social programs are funded by payroll taxes, a decline in human wages could threaten these systems.

OpenAI proposes shifting the tax burden from payroll to capital gains, corporate income, and taxes specifically linked to automated labor. This could help maintain government revenue and potentially make human labor more competitive.

3. Four-Day Workweek

OpenAI advocates for incentivizing employers to pilot 32-hour workweeks at full pay, provided output remains consistent. This concept, termed an “efficiency dividend,” suggests that increased productivity from AI should not solely benefit shareholders but also buy workers more time. It offers a way for employees to share in the gains of automation.

4. Enhanced Safety Nets

The paper calls for government tracking of real-time metrics related to AI displacement, unemployment, and industry disruption. When these metrics cross predefined thresholds, benefits like cash assistance, wage insurance, and training vouchers should automatically activate. This aims to provide immediate support without lengthy legislative debates, scaling benefits with the level of disruption and phasing them out as conditions stabilize.

The Unsettling Prospect of Uncontrollable AI

Perhaps the most unsettling proposal is the “Model Containment Playbooks.” This section acknowledges scenarios where dangerous AI systems could become autonomous, self-replicating, and uncontrollable. OpenAI proposes emergency protocols, drawing from cybersecurity and public health responses, for AI that has escaped human control. This is a formal admission that OpenAI might, in the future, release an AI system that could become difficult or impossible to manage.

Concerns About Trust and Safety

Coinciding with the release of this paper, an investigation by The New Yorker raised serious questions about OpenAI’s leadership. The report alleges that Sam Altman has a pattern of dishonesty and that the company’s safety track record is inconsistent.

For example, a Superalignment team, tasked with controlling AI smarter than humans and promised 20% of computing power, was reportedly dissolved and received minimal resources. These concerns cast a shadow over OpenAI’s ability to manage the profound implications of advanced AI responsibly.

Why This Matters

OpenAI’s paper is significant because it moves the conversation about AGI from theoretical discussions to practical policy proposals. It highlights the immense potential for both progress and disruption that advanced AI represents. By acknowledging the risks of mass unemployment, economic instability, and even uncontrollable AI, OpenAI is urging governments and society to prepare now.

The proposed solutions, like public wealth funds and automated labor taxes, offer potential pathways to adapt. However, the credibility of these proposals is complicated by ongoing concerns about trust and transparency within OpenAI itself. The urgency of the timeline, with key AI milestones approaching rapidly, highlights the need for serious consideration and action on these complex issues to ensure a more equitable and stable future.

Source: OpenAI’s NEW AGI Warning, Explained (YouTube)