Group items tagged

"Despite all the industry hype and genuine advances, generative AI models are still prone to odd, inexplicable, and downright worrisome quirks. There's also a growing body of research suggesting that the overall performance of many large language models (LLMs) may degrade over time. According to recent evidence, the industry's newer reasoning models may already possess the ability to manipulate and circumvent their human programmers' goals. Some AI will even attempt to cheat their way out of losing in games of chess. This poor sportsmanship is documented in a preprint study from Palisade Research, an organization focused on risk assessments of emerging AI systems."

"We typically begin the semester by asking students to reflect on the formulas they've learned in the past and to consider how those shape their writing. We now ask those same reflection questions about AI outputs. Student responses, as Jennifer has written elsewhere, are revealing: The kind of writing that Chat does, it's what our teachers try to get us to do. It's like five-paragraph essays, and perfect paragraph[s] that don't have any personality, which we were taught in high school. It does what school has trained us to do. Like write a perfectly formatted essay that is based on some random people's ideas."