The concept of beauty, as the old saying goes, “is in the eye of the beholder.” Beauty is subjective, opinions varying widely depending on individual tastes and perspectives and interpretations. Similarly, the term Artificial General Intelligence (AGI) is laden with interpretations, making it equally difficult to pin down a universally accepted definition.

What is AGI? A Spectrum of Definitions

AGI, broadly understood, refers to an AI system capable of performing intellectual tasks at or beyond human capability. However, experts have differing takes on what this truly entails:

• Nick Bostrom describes AGI as a system that matches human adaptability, creativity, and self-improvement across intellectual tasks.
• Demis Hassabis envisions AGI as a learning and reasoning system that excels across diverse domains, surpassing humans in efficiency and scale.
• Yann LeCun highlights AGI’s self-supervised learning capabilities, enabling it to understand and interact with the world much like humans.
• Geoffrey Hinton frames AGI as systems capable of human-level reasoning and knowledge application.
• Sam Altman favors a definition of AGI that focuses on outperforming humans at economically valuable work, being capable of generating $100 billion in profits.

These definitions, while insightful, are highly subjective and often tailored to the perspectives of their proponents. They lack the scientific rigor needed for precise analysis, and the landscape is further muddied by tech vendors, AI developers, consultants introducing their own definitions to suit business or strategic narratives.

Toward a Unified Framework: Levels of AGI

In November 2023, a groundbreaking paper, Levels of AGI for Operationalizing Progress on the Path to AGI (arXiv link), sought to bring clarity to this debate. The authors proposed a performance-based categorization of AGI into five distinct levels: Emerging, Competent, Expert, Virtuoso, and Superhuman. This framework aims to provide a structured approach to understanding the evolution of AGI capabilities.

This paper views AGI as a system with capabilities that are comparable to or exceeding those of humans. They exist on a spectrum which can be broken down into five levels.

The Levels of AGI:

1. Emerging AGI: Systems with capabilities comparable to or slightly exceeding those of an unskilled human.
2. Competent AGI: Systems operating at the level of at least the 50th percentile of skilled adults.
3. Expert AGI: Systems that perform at the 90th percentile of skilled adults.
4. Virtuoso AGI: Systems capable of outperforming 99% of humans.
5. Superhuman AGI: Systems that surpass the capabilities of all humans across all tasks.

By articulating these categories, the authors provide a way to assess the current state of AGI and track its progression with precision. As of the paper’s publication, many frontier AI models were categorized as Emerging AGI, demonstrating capabilities that align with or marginally exceed those of unskilled humans. There has been significant progress on AI models since the publication of this framework. With the coming release of OpenAI’s o3 model we may see a capability that meets the definition of a “competent AGI”, another step towards the future.

Defining AGI

Even with this framework, defining AGI involves a bit of subjectivity. When using the term in any conversation it can help reduce ambiguity if you state how you are using the term. The framework presented here can help do that. Here is an articulation of AGI that is contextualized for the business leader:

AGI is an AI system that matches or surpasses human cognitive capabilities across a wide range of cognitive tasks. Specific AGI systems are categorized based on real performance into levels ranging from Emerging to Superhuman. Today’s frontier models meet the criteria for emerging AGI.

Implications for Today and Tomorrow

This structured framework is more than just a theoretical exercise—it has practical implications for businesses, policymakers, and technologists. Understanding AGI’s current capabilities (and limitations) allows stakeholders to make informed decisions about its deployment and future potential. It also offers a lens through which we can evaluate progress, predict disruptive innovations, and prepare for AGI’s impact on industries and society.

The road to AGI is as much about defining milestones as it is about achieving them. With tools like the Levels of AGI framework, we can navigate this complex and rapidly evolving landscape with greater clarity and purpose. For you, whether as a professional, a policymaker, or a curious observer, this means having a clearer understanding of what AGI is today—and what it could become tomorrow.

Related Resources

Decision Intelligence and Establishing an Intelligent Enterprise: The greatest determinant of your success will be the quality of your decisions. We examine frameworks for understanding and reducing risk while enabling opportunities. Topics include Black Swans, Gray Rhinos, Foresight, Strategy, Stratigames, Business Intelligence and Intelligent Enterprises. Leadership in the modern age is also a key topic in this domain.

Artificial Intelligence for Business Advantage.  AI technologies are making continuous advances in domains like industrial robotics, logistics, speech recognition and translation, banking, medicine and advanced scientific research. But in almost every case, the cutting edge AI that drives the advances drops from attention, becoming almost invisible when it becomes part of the overall system.

A Decision-Maker’s Guide to Artificial Intelligence. The megatrend of Artificial Intelligence is transforming the algorithms of business in exciting ways. This reference, aimed at the business decision-maker, will help you make the most of AI in your organization. It provides clear articulations of fundamental concepts, succinct examples of highly impactful use cases, and tips you can put in place to ensure your AI projects stay on track to deliver value. We keep this online reference updated so you will always have access to the best of our thoughts.