By Charles Fadel, Founder & Chairman of the Center for Curriculum Redesign
Artificial General Intelligence (AGI) is widely expected to drastically transform both education and society. From personalised learning to enhanced teacher tools, and the ability to analyse vast amounts of research data, AGI is hyped to make a significant impact on teachers and students alike. As a result, education systems will need to adapt to stay relevant.
However, defining AGI can be tricky. Unlike plain Artificial Intelligence (AI), which is designed for specific tasks, AGI aims to achieve a level of intelligence that is flexible, adaptable, and capable of acting in a manner similar to humans. But the evolving power of AI, which exhibits impressive strengths in some areas while showing glaring weaknesses in others, blurs the line between AI and AGI.
As a result, definitions and perspectives on what AGI actually is vary greatly. While some see AGI as a system that can perform any intellectual task a human can, others believe it should exhibit consciousness or self-awareness. This lack of consensus means that the quest to define AGI remains an ongoing and complex endeavour. The Center for Curriculum Redesign calls for a balanced approach—neither overinflated in ambition nor myopically constrained by existing definitions.
Part of the issue is that people need a more grounded, evolutionary perspective on intelligence, which considers all human competencies across cognitive, affective, and psychomotor domains. These domains are often overlooked in narrow definitions of AGI. Psychomotor tasks, for example, pose unique complexities beyond purely cognitive or LLM (large language model)-centric definitions.
First, let’s explore some definitions of AGI: OpenAI and Google roadmaps, for instance, define certain milestones, but these metrics understandably reflect their respective business priorities rather than a universal gauge of AGI development. Achieving general acceptance of AGI might be another way to define its arrival, albeit a subjective and culture-dependent one.
Another perspective could address AGI’s integration into the labour force and its transformative impact on teaching, nursing, and other professions. Tasks across different domains (e.g., cognitive, affective, and psychomotor) can be categorised by increasing complexity. This may be a far better gauge of AGI’s advances, across a variety of occupational tasks.
There is plenty of skepticism toward LLM-centric views of AGI, noting the inadequacies of such models in handling tasks beyond language processing. While AI has made remarkable progress, including in LLM-based workflows, the leap from narrow AI capabilities to true AGI requires the ability to generalise across fundamentally diverse tasks and domains. In education circles, this is known as Transfer (near, to far), where something learned while young is used decades later in a different context. This leap remains a significant technological barrier to claiming AGI and cannot be achieved on-demand.
Nevertheless, even skeptical AI researchers agree that it is highly likely that some form of AGI will be achieved in the next decade. As a result, education systems cannot be content only to teach AI literacy. They need to pre-emptively consider what else to teach, and do so soonest given their very long decision and implementation cycles, unlike those of technology. Since it is extremely difficult to ascertain AGI’s overall impact, one plausible recommendation is to foster students’ adaptability and self-directed continuous learning, to handle whatever the future holds and stay ahead of AGI. Education systems will need to embed these requirements into their standards, curricula, and lesson plans, which is a tall order. But the conversation needs to start now, and not wait for the AGI steamroller to show up.
