Скачать или смотреть "Open-Source AI is Uniquely Dangerous" – Policy Recommendations from AI Ethics Researcher

In this video we take a look at this article [1] by an AI Ethics researcher claiming that open-source AI is uniquely dangerous.
The author makes a number of policy recommendations that are pretty insane.

[1] https://spectrum.ieee.org/open-source...

Check us out at https://airtrain.ai

#artificialintelligence #regulations #llama #chatgpt #openai #opensource #largelanguagemodels

0:00 Intro
0:39 IEEE
1:00 The author
1:35 General thesis of the article
4:14 Recommendation 1: Pause all new releases
5:08 Recommendation 2: Licensing and registration
6:25 Recommendation 3: Liability for AI developers
7:10 Recommendation 4: Independent audits
7:46 Recommendation 5: Watermarking
8:18 Recommendation 6: Training data
8:59 Recommendation 7: KYC
9:53 Recommendation 9: Incident disclosure