Attention: You are using an outdated browser, device or you do not have the latest version of JavaScript downloaded and so this website may not work as expected. Please download the latest software or switch device to avoid further issues.
14 Aug 2024 | |
The House |
While many advances in LLMs have been hailed as success stories – for instance the ability of Google’s Gemini to generate texts and images – these models have been trained on human-produced content. When such models instead begin to use data that has itself been generated using Artificial Intelligence, quality is compromised. Indeed it can, ultimately, lead to 'model collapse':
'Model collapse is a degenerative process affecting generations of learned generative models, in which the data they generate end up polluting the training set of the next generation.Being trained on polluted data, they then mis-perceive reality.'
Shumailov, I., Shumaylov, Z., Zhao, Y. et al. 'AI models collapse when trained on recursively generated data', Nature 631, 755–759 (2024).
The article is available here.
Professor Yarin Gal is Tutor in Computer Science at Christ Church and Dr Ilia Shumailov was, until recently, a Junior Research Fellow at the college. The article was written in collaboration with PhD student, Zakhar Shumaylov, and Professor Ross Anderson – both of the University of Cambridge – along with Dr Yiren Zhao, Imperial College, London, and Professor Nicolas Papernot, University of Toronto.
The third lecture in the 500th Anniversary Lecture Series was given by The Right Reverend Dr Rowan Williams, former Archbishop of Canterbury, on 11 September in Manchester’s Carol … More...
A new stained glass window has been been installed in the Cathedral, created by the artist Thomas Denny. The window depi… More...
From 18 to 22 August, Christ Church hosted a group of Year 12 students from state schools in its link region, North East… More...
Former Head Librarian, Henry John Richard Wing (known as John), has sadly passed away. John retired from Christ Church i… More...
A team of artificial intelligence experts, including Professor Yarin Gal, has reported a major advance in preventing ope… More...