As an Experienced LLM User, I Actually Don't Use Generative LLMs Often

In the world of artificial intelligence, few tools have sparked as much debate as large language models. Data scientist Max Woolf has been grappling with the ethical implications of these systems, even as he leverages them in his work at BuzzFeed.

"No one in the AI field is happy about prompt engineering," Woolf notes, referring to the art of crafting instructions for AI models. Yet he acknowledges its effectiveness, especially when accessing LLMs through their APIs rather than public interfaces.

Woolf has found LLMs particularly useful for tasks like categorizing articles and generating cluster descriptions. However, he's cautious about their limitations, especially in writing and coding.

"It's hard to be friends with an entity who is trained to be as friendly as possible but also habitually lies due to hallucination," Woolf quips, explaining why he doesn't use LLMs for companionship.