80,000 Hours has updated it's career review on working at an AI company, originally written in June 2023. Here's Claude's summary of the main changes:

Key differences:

1. Tone and stance:
   • The newer article adopts a more cautious and nuanced tone overall.
   • It places greater emphasis on the potential downsides and risks of working at frontier AI companies.
   • The updated version is more explicit about the possibility of contributing to harm.
2. Categorization of companies:
   • The original article refers to "leading AI labs" like OpenAI, Google DeepMind, and Anthropic.
   • The newer article uses the term "frontier AI companies" and includes additional companies like Meta, Mistral, and xAI as contenders.
3. Discussion of specific roles:
   • The updated article provides more detailed information about different types of roles within AI companies and their potential impact.
   • It more clearly distinguishes between roles aimed at reducing catastrophic risks and those that primarily accelerate AI progress.
4. Career capital considerations:
   • While both articles mention career capital benefits, the newer one explores potential downsides more thoroughly, such as the risk of changing one's views due to the work environment.
5. Company responsibility:
   • The updated article places more emphasis on assessing the responsibility of individual companies and provides specific examples of concerns (e.g., events at OpenAI).
6. Alternatives and recommendations:
   • The newer article provides more concrete alternatives to working at frontier AI companies, including government initiatives and non-profit research organizations.
   • It offers more specific advice on how to mitigate downsides if one does choose to work at a frontier AI company.
7. Ethical considerations:
   • The updated version delves deeper into the ethical implications of working on AI capabilities and the potential for contributing to catastrophic risks.
8. Current context:
   • The newer article includes more recent events and developments in the AI field, reflecting the fast-moving nature of the industry.

Overall, the updated article presents a more complex and cautious view of working at frontier AI companies. It maintains that these roles can be high-impact, but it places greater emphasis on the potential risks and the need for careful consideration before taking such positions.

This seems like a significant shift in tone, but there doesn't seem to be any acknowledgement that this is the case