An AI Force would likely push China to create its own. That arms race could turn a technology already slipping out of human control into a weapon.
Articles on LLMs
Displaying all articles
Uncertainty is common in medicine, and AI isn’t very good at navigating it.
The press is an important platform for creating and spreading new linguistic forms.
Urban design depends heavily on human judgment and field-based understanding.
Small language models are like specialised tools in a toolbox, compared to something like ChatGPT that brings the whole workshop.
When a language isn’t in the data, its speakers aren’t in the product – and AI cannot be safe, useful, or fair for them.
Musk’s new WIkipedia rival aims to offer the same service but without the political bias.
In stress-testing AI models, it’s not hard to push them to the brink and make them threaten to harm humans.
AI developers are starting to talk about ‘welfare’ and ‘spirituality’, raising old questions about the inner lives of machines.
It’s hard for ordinary people to distinguish good advice from decisively-voiced bad advice.
Having AI models say how confident they are in their answers could help minimize inaccurate responses. Just don’t be overconfident about their confidence scores.
The rubber met the road for language AIs in 2024. The hard realities led to new, smaller models and safety measures for the big ones. 2024’s R&D also set the stage for the next big thing: AI agents.
‘Dual-use’ AI systems for scientific research might also help nefarious actors create chemical or biological weapons.
Borges imagined an endless library that contained every possible permutation of letters. The truth is out there, but it’s embedded among hordes of lies and gibberish.
As generative language models continue to shape our digital world, the risks of misinformation and misidentification grow ever more pressing.
Looking through the archive of an arts magazine I discovered how Ballard was trying out an early form of computer-generated poetry.
Stereotypes can creep in if ‘culturally sensitive’ robots are not designed with great care.
Unlike the risks from AI’s capabilities, the economic risks from new technologies are knowable and can be mitigated
Future AI large language models like Google’s AMIE might prove to fill gaps in health-care delivery, however, they must be adopted with caution.
Top contributors
-
Research Fellow, Data Science Institute, University of Technology Sydney
-
Head of Digital Economy Research, UCL
-
Researcher, Institute for Innovation and Public Purpose, UCL
-
Associate Professor, School of Computer Science, University of Birmingham
-
Associate Professor, Department of Philosophy, University of Birmingham
-
Visiting Professor of Practice at the UCL Institute for Innovation and Public Purpose, UCL
-
Adjunct professor, Carleton University
-
Professor of Sociology, University of York
-
Professor in the Economics of Innovation and Public Value and Founding Director of the UCL IIPP, UCL
-
Senior Lecturer in Computing and Information Systems, Cardiff Metropolitan University
-
Roger J. Kreuz is a Friend of The Conversation.
Associate Dean and Feinstone Interdisciplinary Research Professor, University of Memphis
-
Associate Professor of Population Health, UNSW
-
2024 Biodefense Fellow at the Council on Strategic Risks and PhD Candidate in Biosecurity at the School of Population Health, UNSW
-
Associate Professor of Computer Science, Director of AMHR Lab, University of South Florida
-
Professor of AI and Behavioral Data Science, University of Technology Sydney