AI Safety & Capabilities News (19th June 2023)
A summary of recent AI news relating to safety and capabilities.
Timestamps:
00:00 - AI News (19th June 2023)
00:29 - UN press conference
01:04 - Anthropic proposal
02:15 - Yampolskiy concerns
03:13 - LeCun perspective
03:54 - Blair/Hague report
05:50 - GPT-4 vs MIT exams
07:23 - Dual use LLMs for biotech
09:28 - Catching errors in scientific papers
10:43 - Fact-checking - Filtir ChatGPT plugin
Corrections:
05:50 - There is some commentary about difficulties reproducing the paper so the results should be taken with a grain of salt ( / 1670053720233750530 )
06:01 - Update: this paper has been withdrawn
Links:
The tweet / 1669405832990040064
UN press conference: • UN Chief on Digital Platforms policy brief...
Anthropic proposal: https://www.anthropic.com/index/chart...
Yampolskiy concerns: https://thehill.com/opinion/technolog...
LeCun perspective: https://www.bbc.com/news/technology-6...
Blair-Hauge report: https://www.institute.global/insights...
GPT-4 vs MIT exams: https://arxiv.org/abs/2306.08997
Dual use LLMs for biotech: https://arxiv.org/abs/2306.03809
Catching errors in scientific papers: https://arxiv.org/abs//2306.07968
Fact-checking - Filtir ChatGPT plugin: https://filtir.com/
For related content:
Twitter: / samuelalbanie
Research lab: https://caml-lab.com/
personal webpage: https://samuelalbanie.com/
YouTube: / @samuelalbanie1
TikTok: / samuelalbanie
Instagram: / samuelalbanie
LinkedIn: / samuel-albanie
(Optional) if you'd like to support the channel:
https://www.buymeacoffee.com/samuelal...
Acknowledgements:
fire animation from "Videezy.com"
Thanks to everyone who provided anonymous feedback on my previous AI News videos! Further feedback is still welcome (the google form is here: https://docs.google.com/forms/d/17P2F...)
#artificialintelligence #AI