Shane Legg proposes "Minimal AGI" is achieved when an AI can perform the cognitive tasks a typical person can. It's not about matching Einstein, but about no longer failing at tasks we'd expect an average human to complete. This sets a more concrete and achievable initial benchmark for the field.
Related Insights
Today's AI models have surpassed the definition of Artificial General Intelligence (AGI) that was commonly accepted by AI researchers just over a decade ago. The debate continues because the goalposts for what constitutes "true" AGI have been moved.
A consortium including leaders from Google and DeepMind has defined AGI as matching the cognitive versatility of a "well-educated adult" across 10 domains. This new framework moves beyond abstract debate, showing a concrete 30-point leap in AGI score from GPT-4 (27%) to a projected GPT-5 (57%).
Shane Legg suggests a two-phase test for "Minimal AGI." First, it must pass a broad suite of tasks that typical humans can do. Second, an adversarial team gets months to probe the AI, looking for any cognitive task a typical person can do that the AI cannot. If they fail to find one, the AI passes.
OpenAI's CEO believes the term "AGI" is ill-defined and its milestone may have passed without fanfare. He proposes focusing on "superintelligence" instead, defining it as an AI that can outperform the best human at complex roles like CEO or president, creating a clearer, more impactful threshold.
The popular concept of AGI as a static, all-knowing entity is flawed. A more realistic and powerful model is one analogous to a 'super intelligent 15-year-old'—a system with a foundational capacity for rapid, continual learning. Deployment would involve this AI learning on the job, not arriving with complete knowledge.
The definition of AGI is a moving goalpost. Scott Wu argues that today's AI meets the standards that would have been considered AGI a decade ago. As technology automates tasks, human work simply moves to a higher level of abstraction, making percentage-based definitions of AGI flawed.
Cutting through abstract definitions, Quora CEO Adam D'Angelo offers a practical benchmark for AGI: an AI that can perform any job a typical human can do remotely. This anchors the concept to tangible economic impact, providing a more useful milestone than philosophical debates on consciousness.
The disconnect between AI's superhuman benchmark scores and its limited economic impact exists because many benchmarks test esoteric problems. The Arc AGI prize instead focuses on tasks that are easy for humans, testing an AI's ability to learn new concepts from few examples—a better proxy for general, applicable intelligence.
Shane Legg, a pioneer in the field, maintains his original 2009 prediction that there is a 50/50 probability of achieving "minimal AGI" by 2028. He defines this as an AI agent capable of performing the cognitive tasks of a typical human.
Current AI models exhibit "jagged intelligence," performing at a PhD level on some tasks but failing at simple ones. Google DeepMind's CEO identifies this inconsistency and lack of reliability as a primary barrier to achieving true, general-purpose AGI.