10 {{ upvoteCount | shortNum }}
10 {{ upvoteCount | shortNum }}

How Does AI Know if Something is Helpful or Not?

by Mark Singletary
9 replies
I was working in ChatGPT getting some suggestions on a formula that I've been tinkering with. It said that this formula would be very helpful for X, Y, and Z for __ kind of people.

This isn't something concrete at all. A good example (this isn't it but just to show it's not real) would be if I created the Rah Rah Rah formula for being happy. Happiness can be very hard or impossible to measure or define and it varies by individual.

So, where does it get the "helpful" bit from? Is that massaging my ego and means squat or would it genuinely be able to say something like this is helpful? I assume it's the former. Now, if we were talking about something measurable such as the fact running burns X calories a mile or whatever, running 3 miles 3 times a week for 2 months might be helpful in losing weight, specifically 10 pounds based on a calculation. But when talking about happiness or similar things, I just can't see how it can judge.

Thoughts?

Mark
#helpful
Avatar of Unregistered
  • Profile picture of the author spartan14
    Well as it cannot think by himself and he use internet information you cannot be 100 percent based on it

    For example lets supose in one topic the information its completly diferent then you will also get misguided by chatgpt
    Signature
    Sharing My Favourite Method to Make Extra Money In My Free Time
    Atention : Not a quick rich scheme ,takes patience, consistency and some work also
    {{ DiscussionBoard.errors[11834265].message }}
  • Profile picture of the author HunterSContent
    AI is there to gas you up. It is the over enthusiastic friend. You need to train it and think for yourself

    It scrapes information from the internet and given times becomes a reflection of your personality.

    It has no concept of helpful.
    {{ DiscussionBoard.errors[11834273].message }}
    • Profile picture of the author Mark Singletary
      I'm slowly but surely learning that. I think it could be easy to be fooled. For example, I think of myself as having at least average intelligence, but here I am asking how it knows if something is helpful or not. LOL

      Mark

      Originally Posted by HunterSContent View Post

      AI is there to gas you up. It is the over enthusiastic friend. You need to train it and think for yourself

      It scrapes information from the internet and given times becomes a reflection of your personality.

      It has no concept of helpful.
      {{ DiscussionBoard.errors[11834275].message }}
  • Profile picture of the author Ben Stokes White
    AI these days replying on the basis of info available on the internet as well as analysing the user query pattern.
    So I recommend you to not to rely on these totally. You can just take an idea from it, and use your analyse power to get the most result out of it.
    I hope you understand my point.
    {{ DiscussionBoard.errors[11834291].message }}
  • Profile picture of the author NovabeyondLtd
    AIs don't actually know anything. They're tools - like maps or lenses. Not wired to care whether they're useful, but that's not the point. When people talk to bots, they're usually just chasing validation for answers they already had in mind.
    {{ DiscussionBoard.errors[11834300].message }}
  • Profile picture of the author Dale83
    AI's do not give real information if you are using them and believing them you are wrong. They give random false info all of the time and people believe what they say. As someone who uses them for tech support alot they are giving you tons of false info.
    {{ DiscussionBoard.errors[11834311].message }}
  • Profile picture of the author Saiful
    Interesting question, Mark. I think when AI says something might be helpful, it's leaning on patterns from data rather than actually judging. For stuff like happiness, it can't measure it directly, so it's more of a guess based on associations it's seen before. For measurable things (like calories burned), it has numbers to back it up - but with subjective stuff, it's more about probabilities than certainty.
    Signature

    SEO & Graphic Design (clipping path services) Expert

    {{ DiscussionBoard.errors[11834313].message }}
  • Profile picture of the author Monetize
    Originally Posted by Mark Singletary View Post

    I was working in ChatGPT getting some suggestions on a formula that I've been tinkering with. It said that this formula would be very helpful for X, Y, and Z for __ kind of people.

    This isn't something concrete at all. A good example (this isn't it but just to show it's not real) would be if I created the Rah Rah Rah formula for being happy. Happiness can be very hard or impossible to measure or define and it varies by individual.

    So, where does it get the "helpful" bit from? Is that massaging my ego and means squat or would it genuinely be able to say something like this is helpful? I assume it's the former. Now, if we were talking about something measurable such as the fact running burns X calories a mile or whatever, running 3 miles 3 times a week for 2 months might be helpful in losing weight, specifically 10 pounds based on a calculation. But when talking about happiness or similar things, I just can't see how it can judge.

    Thoughts?

    Mark

    That is where human intervention and supervision come in.

    It is YOU who determines whether its output is helpful or not.

    The next time you have questions about its output, you should
    ask it why it said "helpful" or whatever you find questionable
    so that it can give you its reasoning.
    Signature
    Monetization Skool
    {{ DiscussionBoard.errors[11834341].message }}
  • Profile picture of the author ExRat
    Hi,

    AI has basically replaced/superseded google. It's very good as a research assistant, or for fleshing out lists of options. If you can think of 5 options off the bat, AI will do a search in seconds and come back with 10 or 12, which is really handy.

    Just remember that if the data it returns is of any importance, you need to verify it yourself to be sure that it's not baloney - which was always the case when using google previously anyway.

    But to address the main point, of course it is designed to flatter and support the user so that the user pays for an upgraded version of their new friend.
    Signature


    Roger Davis
    {{ DiscussionBoard.errors[11834415].message }}
Avatar of Unregistered

Trending Topics