Large Language Models

Thanks to John Naughton’s Guardian column (https://www.theguardian.com/commentisfree/2023/jan/07/chatgpt-bot-excel-ai-chatbot-tech) I picked up this illuminating paper – a clear and confident perspective on how LLMs like GPT work (“next token predictors”!).

Download here: Murray Shanahan, 2022, ‘Talking about Large Language Models’, https://arxiv.org/abs/2212.03551

Detecting the non-human

Sniffing out GPT

An app that measures “burstiness” and “perplexity” to detect a GPT source.

However,

” … it does not work well with essays written by good writers. It false flagged so many essays as AI-written.”

https://www.theguardian.com/technology/2023/jan/12/college-student-claims-app-can-detect-essays-written-by-chatbot-chatgpt

https://www.bbc.co.uk/news/world-us-canada-64252570

Tests are dead?

“…for educators, the technology opens the door to widespread cheating on homework and take-home assignments, and many have been scrambling to rethink the nature of assessment or otherwise discourage students using the tool.”

“… ChatGPT is capable of passing — or at least nearly passing — the US medical licensing exams, according to researchers in the US who put it to the test.”

https://www.notion.so/ChatGPT-appears-to-pass-medical-school-exams-Educators-are-now-rethinking-assessments-ABC-News-7668227901fc4b72b9ed369a7bd4a363