| Website | https://www.lilianedwards.co.uk/ |
| University site | https://www.ncl.ac.uk/law/people/profile/lilianedwards.html |
| Wikipedia | https://en.m.wikipedia.org/wiki/Lilian_Edwards |
| Pronouns | She/her |
Pleased to note my Cambridge lecture on how technical approaches to the right of explanation have gone haywire in the LLM era, is already up!
I discuss the spread of legal mandates of explanations of algorithmic decisions but at same time increasing unreliability of explanations offered by "reasoning" LLMs. Chain of Thought not = explainability!!
Many thanks to @CIPIL who um may not be here? And to @jennifercobbe for her v kind invitation!
Not only can #preprints be cited, citations to preprints are growing. Using the literature on #AI policy for a case study, these authors found that citations of preprints increased 8x over the last decade, from 5% to 40%.
https://arxiv.org/abs/2510.16477
This study of literature focusing on 'AI Policy' over the past decade, found that citations of preprints, publications on platforms such as arXiv, have increased from five percent to forty percent across three major regions: the U.S., U.K. & E.U., and South Korea. We compare regional responses of preprint citations across the global disruptions of COVID-19 and the release of ChatGPT. We discuss driving factors and risks of preprint normalization, which follows the trend in computer science.
The NHS Ten Year Plan states: "One study found AI helped improve performance of highly skilled workers by 40%, in certain tasks achievable by AI, compared to those who didn’t use it"
The study it cites tested management consultants advising CEOs on stuff like developing product and marketing strategies and producing reports on market segments. "ChatGPT lets management consultants do their job more efficiently" is perhaps not generalisable to other professions and fields.
petersuber
Tom Morris