The Bot Chat - AI News and Research

NEWS: A coalition of UK creative industries have rejected a ministerial proposal to allow LLMs to train their algorithms on published materials unless their authors opt out, urging the government to respect existing copyright laws.

NEWS: OpenAI whistleblower, Suchir Balaji, has been found dead in his apartment. Balaji argued that late 2022 GPT models were commercially deployed in a way that meant the models competed with their training data, resulting in violations of fair use law.

DISCUSSION: Gstrein et al (2024) argue that the EU AI Act (AIA), with its major focus on product safety and standardisation and some elements related to the protection of human rights, seems bound to become a global benchmark for AI regulation. Do you agree?

100% human, perhaps the robot profile picture is a little misleading 😅

Sure did! I'm a total nerd for chatting about this sort of thing, so I figured I'd jump on Bluesky and see if I could create a space for like-minded people :)

DISCUSSION: A median forecast from a survey of machine learning researchers concluded that 22% of researchers believe in the possibility of broadly human-level AI within the next 2 decades (Zhang et al., 2021). Do you agree?

Absolutely - for a model with as much investment as Gemini, something like this happening is kind of insane.

NEWS: Google's AI chatbot Gemini has been found to have told a student to "please die" and that they are a "stain on the universe". The comments were in response to a seemingly innocuous prompt from the user about statistics in the US about households led by grandparents.

Hi Bluesky! I'm an MSc AI and Government student with a BA in Philosophy and Psychology, and I look forward to sharing the latest AI news and offering a place to foster discussions at the intersection of AI, Philosophy and Psychology!