7 Quotes about the Limitations of AI

While AI can enhance individual creativity, it might do so at the expense of collective diversity and novelty in creative works. PsyPost

The AI programs aren’t necessarily doing something no human can; they’re doing something no human can in such a short period of time. Sometimes that’s great, as when an AI model quickly solves a scientific challenge that would have taken a researcher years. Sometimes that’s terrifying, as when (they appear) capable of replacing entire production studios. -The Atlantic

“On average 30% of the time the AI models spread misinformation when asked about claims in the news. On average 29% of the time, the AI models simply refused to respond to prompts about false claims in the news over the past month. Instead, the models delivered only non-responsive responses.” -News Guard

While AI models are starting to replicate musical patterns, it is the breaking of rules that tends to produce era-defining songs. Algorithms ‘are great at fulfilling expectations but not good at subverting them, but that’s what often makes the best music,’ Eric Drott, a music-theory professor at the University of Texas at Austin.” How can we be more human than an AI? Produce creative work that goes beyond the expected, the predictable, the established and popular. -The Atlantic

Recent brain scans suggest we don’t need language to think. A group of neuroscientists now argue that our words are primarily for communicating, not for reasoning. "Separating thought and language could help explain why AI systems like ChatGPT are so good at some tasks and so bad at others. These programs mimic the language network in the human brain — but fall short on reasoning." - New York Times

If an LLM can be trained on 17th-century texts, it can just as easily be trained on QAnon forums, or a dataset that presupposes the superiority of one religion or political system. Use a deeply skewed bubble machine like that to try to understand a book, a movie, or someone's medical records and the results will be inherently biased against whatever — or whoever — got left out of the training material. -Business Insider

At times, A.I. chatbots have stumbled with simple arithmetic and math word problems that require multiple steps to reach a solution, something recently documented by some technology reviewers. The A.I.’s proficiency is getting better, but it remains a shortcoming. -New York Times