The leaders of the artificial intelligence industry are increasingly vocal about the transformative, and sometimes apocalyptic, potential of their creations. This comes as their companies secure substantial funding, often for infrastructure like data centers, while simultaneously issuing warnings about existential risks.
A running tally has emerged to track the pronouncements of AI executives, noting shifting timelines and evolving justifications for the pace of development. The log aims to provide clarity on the often-conflicting statements and the moving goalposts set by industry leaders.
The Promise and Peril of AGI
Artificial General Intelligence (AGI), defined as AI that matches or surpasses human intelligence, is frequently presented as a developmental milestone. OpenAI describes AGI as a technology that could "elevate humanity by increasing abundance, turbocharging the global economy, and aiding in the discovery of new scientific knowledge." The potential benefits include granting "incredible new capabilities" and acting as a "great force multiplier for human ingenuity and creativity." However, OpenAI also acknowledges "serious risk of misuse, drastic accidents, and societal disruption," emphasizing the need to "get it right."
Shifting Timelines for AGI
Predictions for when AGI might arrive have varied significantly and frequently shifted:
- In September 2026, Jensen Huang suggested that "AGI has arrived" with the GPT-6 Astra model.
- Sam Altman stated in August 2026 that OpenAI would have an internal AGI system by the end of that year.
- Elon Musk predicted in December 2025 that xAI could achieve AGI within a few years, potentially as soon as 2026.
- Google's Demis Hassabis indicated in January 2025 that AGI was likely "three to five years away."
- Altman also expressed in November 2024 that AGI was expected in 2025.
- In April 2024, Elon Musk forecasted that AI would be smarter than the smartest human within two years.
- Nvidia's Jensen Huang stated in March 2024 that AGI could be achieved within five years, based on specific performance tests.
- As far back as October 2023, Google's Blaise Agüera y Arcas and Stanford's Peter Norvig declared that "Artificial General Intelligence is already here."
Evolving Terminology and Concerns
Recent pronouncements have introduced new terminology to describe AI's advancement:
-
Recursive Self-Improvement: This concept, gaining traction in June 2026 following Anthropic's discussion on "When AI builds itself," refers to AI systems capable of autonomously developing and improving subsequent versions of the technology. Anthropic noted that while "not there yet" and "not inevitable," this trend could accelerate development faster than many institutions anticipate.
-
Pacing the Frontier: Introduced in July 2026 through a statement by 1,386 AI employees, this call advocates for slowing AI development. The signatories requested government support for international efforts to create the necessary technical and governance tools to "deliberately pace the frontier of automated AI development."
-
Concerns about AI's immediate impact have also surfaced. In September 2026, a statement via The New York Times likened the necessary pain of AI development to "surgery" requiring "enormous amount of pain and suffering" to save humanity, suggesting a continued reliance on fossil fuels might be a prerequisite for sustainable energy transition.
-
Anthropic's CEO, in September 2026, expressed a "greater than 10%" chance of AI killing all humans within the next decade, admitting that the company "do[es] not yet have a plan to solve alignment for superintelligence."