Artificial intelligence (AI) has made significant strides in areas such as mathematics, exemplified by systems developed by Google and OpenAI that performed impressively at the International Math Olympiad. However, AI still struggles with basic common-sense tasks, a discrepancy that researchers refer to as 'jagged intelligence.' This term, coined by Andrej Karpathy, reflects the uneven performance of AI, which excels in some domains while failing in others. For instance, when asked whether to walk or drive to a repair shop just 50 meters away, leading AI systems provided incorrect advice, highlighting their limitations in practical reasoning.

The implications of jagged intelligence extend to the labor market, as economists and researchers analyze how AI's varied capabilities may affect employment. While entry-level programming jobs may be at risk, the overall impact on other professions remains uncertain. Experts suggest that AI could enhance human productivity rather than replace workers entirely, similar to how calculators improved efficiency without displacing accountants.

Recent tests, such as the ARC-AGI 3 benchmark, have shown that leading AI systems struggle with problem-solving tasks that an average person can easily handle. This indicates that while AI can learn specific skills through pattern recognition and reinforcement learning, it lacks general intelligence and the ability to innovate or plan ahead. As AI technology continues to advance, understanding its jagged intelligence will be vital for anticipating its future role in the workforce.