news.volyx.in

Elon Musk sues Sam Altman, Greg Brockman, and OpenAI [pdf] (courthousenews.com)

1462 points by modeless · 899 days ago · 1208 comments on HN

Article summary

Elon Musk is suing Sam Altman, Greg Brockman, and OpenAI, reportedly over the definition of Artificial General Intelligence (AGI) and its implications on licensing agreements with Microsoft. The lawsuit is related to Microsoft's exclusive licensing of OpenAI's GPT-3 language model and its claims to GPT-4, which it asserts has not reached the level of AGI. The definition of AGI and its potential attainment by GPT-4 are central to the dispute. The lawsuit's details are not publicly available, but the issue has sparked a discussion on the definition and implications of AGI.

Main themes

  • Artificial General Intelligence
  • Licensing agreements
  • Microsoft and OpenAI partnership
  • GPT-3 and GPT-4 language models
  • Definition of AGI
  • Implications of AGI attainment

What commenters say

  • The definition of AGI is unclear and may be subjective, making it difficult to determine when it has been attained.
  • GPT-4 may not be considered AGI, as it does not outperform humans at most economically valuable work, and its capabilities are still limited.
  • The concept of AGI is flawed, and the focus on it is a distraction from more practical and incremental progress in AI development.
  • The possibility of achieving AGI through current technology, such as transformer models, is considered low or zero by some, while others believe it is non-zero but not deserving of the current hype.
  • The attainment of AGI would have significant implications for licensing agreements and the control of AI technology, particularly in the context of the Microsoft-OpenAI partnership.
  • The development of AI should focus on starting from simple sensory input rather than relying on statistical pattern matching and large language models.
  • The progress towards AGI is exponential, and the amount of money invested in deep learning infrastructure is increasing rapidly, which could lead to significant advancements in AI capabilities.
  • The idea that AGI can be achieved through current architectures, such as transformer models, is unlikely due to their limitations, such as the lack of true recursion.