Entangled States

A study at MIT has found that Generative AI tools are having a significant impact on the way students are learning. That’s not terribly surprising in of itself, but the report presents data and results that support the assertion and makes suggestions about what to do about it. Skipping to the end of the report, in summary: Adapt educational processes for an AI-aware world (Section 3.1) Campus-wide and as soon as possible, review what our […]

AI-generated bot swarms pose a significant threat to democracy by making fringe views appear mainstream through manipulated social evidence. These bots can compromise training data for future AI models, exacerbating misinformation. While banning isn't a solution, changing the economic incentives behind manipulation is crucial to safeguarding democratic integrity.

The more I think about what Chatbots are and what they aren’t, the more I think of them, not so much as a stochastic parrot, but as a tool that we’re trying to use for the wrong purpose. There are some discrete tasks, in particular fields that Chatbots can help with (summarization, coding, some well understood math problems) but there are others where they can’t. The limit is because of what they are, not what […]

People Are Being Involuntarily Committed, Jailed After Spiraling Into “ChatGPT Psychosis”: Jared Moore, the lead author on the Stanford study about therapist chatbots and a PhD candidate at Stanford, said chatbot sycophancy — their penchant to be agreeable and flattering, essentially, even when they probably shouldn’t — is central to his hypothesis about why ChatGPT and other large language model-powered chatbots so frequently reinforce delusions and provide inappropriate responses to people in crisis. […] “There’s […]