• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
  • HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI

    Arvind Narayanan on Upholding AI Research Ethics

    Arvind Narayanan argues distinctions remain possible by tracking model evaluation calls versus research assistant uses and funding sources.

    AN
    GI
    2 Sources, 30d ago, first seen 30d ago

    TLDR

    Princeton professor Arvind Narayanan replied that moral distinctions can be upheld even when using AI tools as research staff. He stated the difference holds if API calls are tracked separately for evaluating the invoked model versus using models as assistants, along with separate accounting for what pays for each. Narayanan noted there will always be a moral difference between company-funded arrangements and other setups. The reply appears in an AI topic discussion and includes no further details on specific cases or outcomes.

    Combined views

    183

    2 Sources, first seen 30d ago

    Combined views

    183

    2 Sources, first seen 30d ago

    5 likes
    5 likes
    2 comments

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    2 comments
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    2 Sources

    @random_walkerEven in a world with AI staff this distinction can be meaningfully upheld as long as we track which API calls are for evaluating the model being invoked and which calls are for models being used as research assistants, and what pays for what. Surely there will always be a moral difference between "company funds our salaries" and "company enables us to test their models".
    @geoffreyirving@RishiBommasani @METR_Evals I think it's harder to separate the influence of these two modes of compute in our case, given that they are both inputs into the core "do alignment research" work, and mostly not as an input to particular lab collaborations (a la METR token use as part of evals).

    2 Sources

    @random_walkerEven in a world with AI staff this distinction can be meaningfully upheld as long as we track which API calls are for evaluating the model being invoked and which calls are for models being used as research assistants, and what pays for what. Surely there will always be a moral difference between "company funds our salaries" and "company enables us to test their models".
    @geoffreyirving@RishiBommasani @METR_Evals I think it's harder to separate the influence of these two modes of compute in our case, given that they are both inputs into the core "do alignment research" work, and mostly not as an input to particular lab collaborations (a la METR token use as part of evals).