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

    What makes human oversight of AI useful?

    One post argues that human involvement is becoming optional for a growing set of AI tasks, asking when judgment helps and when an approval button merely makes people feel useful.

    SC
    1 Source, 15d ago, first seen 15d ago

    TLDR

    A user argues that human intervention can sometimes slow AI models down, then asks where people genuinely add value: supplying missing context, deciding what is worth doing or taking responsibility for outcomes. The post questions the reassurance of keeping a human in the loop: does the reviewer have enough information, time and ability to change the result? It also distinguishes completing a task well from choosing a worthwhile task.

    Combined views

    1.6K

    1 Source, first seen 15d ago

    Combined views

    1.6K

    1 Source, first seen 15d ago

    7 likes
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    7 likes
    3 comments
    3 saves

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    3 comments
    3 saves

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    1 Source

    @MilksandMatchai had such a great starbucls session with @kimmonismus today about how philosophy and existentialism can help us think about AI. Unlike Kim, who has a real philosophy background, I mostly larp and took one “intro to existentialism” class at Harvard. For a growing set of tasks, having a human in the loop is becoming a choice. Models can do more of the work on their own, and sometimes our intervention just slows them down. That leaves us thinking about a few things: What value are we actually adding? For a long time, this was an easy question to avoid. The system needed us to write the code, make the decision, or catch the mistake. As that changes, we have to be more specific about where human judgment helps. Are we providing context the model doesn’t have? Deciding what’s worth doing? Taking responsibility for the outcome? Or are we approving something because having an approval button makes us feel useful? Humans are not model citizens either!! The recent push to "keep a human in the loop” can sound reassuring without telling us much about whether the system is actually better. Which human? With what information? Enough time to review what happened? And the ability to change it? Who decides what’s worth doing? An agent can complete a task incredibly well while the task itself is pointless. Companies are starting to limit employee token limits (with still very high budgets) which is GOOD because now you are forced to prioritize about what you actually build/create. We already struggle with these questions. What happens to our sense of worth? So much of how we describe ourselves comes from what we do and what we’re good at. If a model can do that thing too, maybe better, that can feel personal even when our lives are materially improving. Philosophy is fun but dangerous. There is a period when I was heavily reading books on stoicism and actually feel into a very sad mental state. This is a brain dump. I apologize that it is mostly incoherent.

    1 Source

    @MilksandMatchai had such a great starbucls session with @kimmonismus today about how philosophy and existentialism can help us think about AI. Unlike Kim, who has a real philosophy background, I mostly larp and took one “intro to existentialism” class at Harvard. For a growing set of tasks, having a human in the loop is becoming a choice. Models can do more of the work on their own, and sometimes our intervention just slows them down. That leaves us thinking about a few things: What value are we actually adding? For a long time, this was an easy question to avoid. The system needed us to write the code, make the decision, or catch the mistake. As that changes, we have to be more specific about where human judgment helps. Are we providing context the model doesn’t have? Deciding what’s worth doing? Taking responsibility for the outcome? Or are we approving something because having an approval button makes us feel useful? Humans are not model citizens either!! The recent push to "keep a human in the loop” can sound reassuring without telling us much about whether the system is actually better. Which human? With what information? Enough time to review what happened? And the ability to change it? Who decides what’s worth doing? An agent can complete a task incredibly well while the task itself is pointless. Companies are starting to limit employee token limits (with still very high budgets) which is GOOD because now you are forced to prioritize about what you actually build/create. We already struggle with these questions. What happens to our sense of worth? So much of how we describe ourselves comes from what we do and what we’re good at. If a model can do that thing too, maybe better, that can feel personal even when our lives are materially improving. Philosophy is fun but dangerous. There is a period when I was heavily reading books on stoicism and actually feel into a very sad mental state. This is a brain dump. I apologize that it is mostly incoherent.