Artificial general intelligence will be born from a burst of laughter

Artificial intelligence, while rapidly advancing, still struggles with nuanced human expression, particularly humor. A recent test involving ChatGPT's jokes revealed its limitations, eliciting polite, rather than genuine, amusement.

In 2024, researchers conducted blind tests where participants rated jokes from ChatGPT against those made by humans. While the AI often produced technically correct jokes, it failed to impress professional comedians. When tested by 20 humor experts, AI-generated jokes were described as reminiscent of "1950s cruise ship jokes, only slightly less racist."

“

Why do divers always dive backward? Because otherwise, they fall into the boat.

This assessment highlights a significant gap between AI's capacity for pattern recognition and its ability to grasp the subtleties of comedic timing, cultural context, and genuine wit that resonate with audiences. The current state of AI, as exemplified by ChatGPT, can mimic joke structures but lacks the creative spark and originality that defines effective humor.

The author posits that the development of true artificial general intelligence (AGI) might not follow a predictable path. Instead, it could arise from unexpected sources, perhaps even from the seemingly trivial act of laughter itself. This perspective challenges conventional thinking about AI development, suggesting that breakthroughs may come from unconventional areas, far removed from purely logical or data-driven processes.

“

jokes from the 1950s cruise ship, only slightly less racist.