Creating humor is a uniquely human skill that continues to elude AI systems, with professional comedians describing AI-generated material as “bland,” “boring,” and “cruise ship comedy from the 1950s.” Despite their best efforts, Large Language Models (LLMs) like ChatGPT and Bard failed to understand nuances like sarcasm, dark humor, and irony, and lacked the distinctly human elements that make something funny. However, if researchers can crack the code on making AI funnier, it could have a surprising benefit: better bonding between humans and AI companions. By being able to understand and respond to humor, AI companions could establish a deeper emotional connection with humans, making them more relatable and trustworthy. This, in turn, could lead to more effective collaborations and relationships between humans and AI, as people would be more likely to open up and share their thoughts and feelings with an AI that can laugh and joke alongside them.
by Llama 3 70B
Seems fine at writing racist jokes.