Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

The Download: how OpenAI tests its models, and the ethics of uterus transplants

Nov 22, 2024 - technologyreview.com
OpenAI has revealed some of its safety-testing processes for its large language models, aiming to identify potential harmful or unwanted behavior. The company uses an extensive network of human testers to vet the behavior of its models before release, and is developing ways to automate parts of the testing process. In other news, over 135 uterus transplants have been performed globally in the last decade, raising legal and ethical questions about who should be offered the procedure and who should pay for it.

In technology news, OpenAI may launch a web browser, posing a challenge to Google. Border militias are prepared to assist with Trump's deportation plans, Russia has used a new type of missile against Ukraine, and Microsoft is about to turn 50. China has surpassed Germany in industrial robot adoption, and a French startup plans to make fuel out of thin air. WhatsApp plans to start transcribing voice messages, and second-hand phones are being promoted as a more sustainable choice.

Key takeaways:

  • OpenAI has released two papers detailing its safety-testing processes for its large language models, using a method known as red-teaming.
  • Over 135 uterus transplants have been performed globally in the last decade, raising legal and ethical questions about who should be offered the procedure and who should pay for it.
  • OpenAI may launch a web browser, which could potentially challenge Google.
  • Alán Aspuru-Guzik, a chemist, is using AI and automation to speed up the process of materials discovery, which could help address challenges posed by climate change.
View Full Article

Comments (0)

Be the first to comment!