Google DeepMind CEO Demis Hassabis stated that general artificial intelligence may emerge around 2030. He believes that AI is no longer a distant scientific research goal, and society has little time left to prepare.
2026 is considered a turning point.
Hassabis recently stated at an event at Stanford University's Graduate School of Business that AGI (Advanced Genetic Intelligence) may arrive before the end of this century, roughly around 2030, within a year. He described this phase as the beginning of a new era.
He also mentioned that 2026 will be a significant turning point. As AI agents and tool invocation capabilities begin to truly enter the workplace, developers will have a clearer understanding of what steps are still needed to achieve AGI (AI-enabled gadgets).
Industry timelines are still not unified
There is currently no consensus within the AI industry regarding when AGI will be realized. OpenAI CEO Sam Altman previously stated that the company already knows how to build AGI in the traditional sense and expects AI beings to gradually enter the labor market.
Anthropic CEO Dario Amodei and Elon Musk have both given rather aggressive timelines. Musk stated last December that AGI could emerge in 2026, and that by 2030, the overall intelligence of AI could surpass that of all humans.
Disagreements remain regarding definition and testing.
However, some argue that the current model is still significantly different from true general intelligence. Part of the controversy stems from the definition itself: what constitutes AGI (Advanced Generative Intelligence) is not a unified standard within the industry.
In March of this year, the ARC Prize Foundation released the ARC-AGI-3 benchmark test to examine the learning and adaptation capabilities of AI systems in unfamiliar environments. Leading models from Google, OpenAI, Anthropic, and xAI all scored below the top 1%, while human participants achieved perfect scores.
This leaves the AGI discussion stuck on two lines: on the one hand, leading companies continue to emphasize that technological progress is accelerating; on the other hand, existing test results show that the current system is still far from achieving general reasoning capabilities at the human level.
