☣️ AI Outsmarts Virus Experts, Raising Biohazard Fears
A study from the Center for AI Safety and SecureBio shows top chatbots now beat PhD virologists at lab problem-solving, raising both hopes for faster research and fears of misuse.
On the 322-question Virology Capabilities Test, OpenAI's o3 answered 43.8 % correctly, nearly doubling the human average of 22.1 %. Moreover, o3 performed better on the test than 95% of test-takers; Claude 3.5 Sonnet beats 75% of humans.
Because these skills could be weaponized, the authors urge limiting public access and adding strict "know-your-customer" checks.
Read More
Post #441
200K
- 👍 196
- ❤ 99
- 🥰 32
- 🔥 20
- 😁 10
- 🎃 10
- 😨 10
- 👾 9
- 👏 8
- 🤣 7
- 🤔 2