Alt Text

The latest artificial intelligence benchmark, ARC-AGI-2, has sparked controversy by introducing its own test to measure human intelligence, despite being unable to perform basic social functions like pretending to laugh at unfunny jokes.

The AI system, which can solve complex mathematical equations in nanoseconds, created a 47-point assessment focusing exclusively on uniquely human skills such as “efficient time-wasting” and “convincing your boss you’re working while watching YouTube.”

“While the AI can process terabytes of data instantly, it still can’t figure out why humans need to sleep or why we enjoy watching cat videos,” explained Dr. Sarah Neural, Head of Human Superiority Studies. “It’s like having a genius student who can’t tie their shoes.”

The benchmark includes tests for “strategic meeting avoidance” and “optimal excuse generation,” inspired by the viral “lazy girl job” trend. “Our latest results show that 98% of humans can successfully pretend to work while actually browsing social media. The AI keeps accidentally solving problems instead,” reported Bob Binary, Chief of Human Testing.

The AI is currently taking a processing power nap to analyze the data day tasks, though researchers suspect it’s just avoiding another awkward conversation about the weather.

AInspired by: ARC-AGI-2 Benchmark Released