Ex-OpenAI Researcher Predicts AGI by 2027

Leopold Aschenbrenner, a former safety researcher at OpenAI, known for developing ChatGPT, has recently expanded on the topic of artificial general intelligence (AGI) in his new essay series on the future of artificial intelligence (AI). The series, titled “Situational Awareness,” provides a detailed look at the current state and potential advancements in AI systems over the next decade. Compiled in a 165-page PDF, the series of essays was updated on June 4th.

In his essays, Aschenbrenner delves deeply into AGI, a type of AI designed to match or exceed human abilities in a broad spectrum of tasks. He points out that AGI stands apart from other types of AI, such as artificial narrow intelligence (ANI) and artificial superintelligence (ASI). According to Aschenbrenner, the arrival of AGI by 2027 is a plausible scenario. He forecasts that AGI systems will surpass the intelligence of college graduates by 2025 or 2026. He asserts that by the end of the decade, AGI entities will be smarter than humans, achieving what he terms “superintelligence.” He anticipates significant national security impacts not seen in fifty years.

Aschenbrenner also speculates that future AI systems will potentially have intellectual capacities comparable to a seasoned computer scientist’s. He boldly predicts that AI labs will manage to train general-purpose language models at astonishing speeds; for instance, a model on par with GPT-4 could be trained in just one minute by 2027.

Highlighting the inevitability of AGI, Aschenbrenner urges the AI community to confront the forthcoming challenges and opportunities. He explains that the leading minds in AI have aligned themselves with a concept he calls “AGI realism.” This viewpoint rests on three core principles that are interlinked with the national security and technological advancement in the United States.

The release of Aschenbrenner’s essay series follows a controversial period in his career. He was reportedly dismissed from OpenAI for allegedly leaking confidential information. He was an associate of OpenAI’s chief scientist, Ilya Sutskever, who attempted but failed to remove OpenAI’s CEO, Sam Altman, in 2023. Aschenbrenner has dedicated his latest series to Sutskever.

In addition to his writings, Aschenbrenner has recently ventured into the investment world. He has founded an investment firm that focuses on AGI, boasting significant backing from prominent investors, including Stripe CEO Patrick Collison.

Through his essays, Aschenbrenner paints a picture of a transformative future shaped by AGI. He speculates on its societal impacts, hinting at both the promise and peril that such advancements could bring. The researcher suggests that the world is on the brink of a significant shift, driven by exponential advancements in AI technology.

As Aschenbrenner’s predictions unfold, the discourse around AGI will likely intensify. Whether his forecasts come to pass or not, his essays contribute to the ongoing debate about the future of artificial intelligence and its implications for humanity. By envisioning a future where AI systems have capabilities that far exceed our own, Aschenbrenner challenges us to think critically about the ethical, social, and political dimensions of such developments.