From AI Hallucinations to Befriending Chatbots: AI Questions, Answered | WSJ Tech News Briefing

แชร์
ฝัง
  • เผยแพร่เมื่อ 21 พ.ย. 2024

ความคิดเห็น • 53

  • @esm2000
    @esm2000 ปีที่แล้ว +6

    I loved this format. taking questions from viewers is something you guys should do more of

  • @rsaunders57
    @rsaunders57 ปีที่แล้ว +8

    The easiest test for generative AI is to add "cite your sources" to the end of your question. With a human you get a citation, which might be bad but at least it's a citation. The generative AIs respond with "According to my sources" followed by their ordinary answer.

    • @mateuszbugaj799
      @mateuszbugaj799 ปีที่แล้ว +11

      Bing provides sources most of the time, and ChatGPT will soon have the ability to search the internet and provide references as well.

    • @josephlynch7655
      @josephlynch7655 ปีที่แล้ว

      Try Bing. It always cites sources.

    • @AnnasVirtual
      @AnnasVirtual ปีที่แล้ว

      it will just going to made up sources because it just doesn't actively go to the internet to assemble texts (at least the one that is not connected to the internet), it just predicting the next word in a sentence

    • @kamu747
      @kamu747 ปีที่แล้ว +4

      Just keep in mind that their citations can some times be made up. So be keen.
      Bing seems to be doing a great job at it. Still, the responsibility lies with you to cross-check that the information is indeed in the sources because once again, it can sometimes get the details wrong.

    • @amatvkhmer
      @amatvkhmer ปีที่แล้ว

      ai makes work very easy

  • @siddharthsood86
    @siddharthsood86 ปีที่แล้ว +5

    This one was a good episode. Keep it up wsj

  • @randymarsh6666
    @randymarsh6666 ปีที่แล้ว +3

    I'm john Conner. This must stop!!!!

    • @JuraKosednar
      @JuraKosednar ปีที่แล้ว

      Watch Matrix first John...

  • @TheAugustus24
    @TheAugustus24 ปีที่แล้ว +1

    This episode was great, thank you.

  • @adolfolopezmartinez3696
    @adolfolopezmartinez3696 ปีที่แล้ว +1

    Back to school with WSJ.

  • @vit3869
    @vit3869 ปีที่แล้ว +1

    So very interesting.

  • @SoCalFreelance
    @SoCalFreelance ปีที่แล้ว +1

    The movie Her (2013) provides some thought-provoking ideas on the potential future of AI companionship.

  • @SaurabhOKumar
    @SaurabhOKumar ปีที่แล้ว +1

    Why you re-tuned the program's commencing music Zoe??... The earlier original music was so chilling to ears as it actually created more focus, enthusiasm and interest in further listening your program head-on, lady!! 😮

  • @dylan_curious
    @dylan_curious ปีที่แล้ว +1

    OMG this is so interesting! I never thought about AI being used for emotional or friendship purposes. It would be cool to have an AI friend to talk to when you're feeling lonely or need advice. But at the same time, I wonder if it's kind of weird to have a relationship with a machine, you know? And how long would it take to actually develop a bond with an AI? I guess it depends on how advanced the technology is and how realistic the AI can simulate emotions and conversation. It's definitely something to think about as AI becomes more and more integrated into our daily lives

  • @Brian_Chui
    @Brian_Chui ปีที่แล้ว +1

    Great episode. Only one thing, the speaker lost me when she said that she believes China is heavily regulated in AI so we don’t have to worry about it, as if China has any regulations there can be trusted. If this is the mindset for our experts and regulators, we will then be far behind in this race. This tool was invented in the states, but unfortunately will be exploited elsewhere. This in my mind will come much sooner than singularity folks.

  • @NikoKun
    @NikoKun ปีที่แล้ว +1

    The "autocomplete" explanation at 2:30 is okay for describing models from a couple years ago, or giving people a very over-simplified idea of how all this started.. But that explanation cannot explain the level of reasoning the latest LLMs display, as well as the many other emergent behaviors we're beginning to see.

  • @amatvkhmer
    @amatvkhmer ปีที่แล้ว +1

    ai makes work very easy

  • @potatomatop9326
    @potatomatop9326 ปีที่แล้ว +5

    Isn't befriending Chatbots like talking to yourself?

    • @johndank2209
      @johndank2209 ปีที่แล้ว

      No, you are interacting with something that isn't you.

    • @pixelfairy
      @pixelfairy ปีที่แล้ว

      It is if you train it on your own text.

    • @1000xtati
      @1000xtati ปีที่แล้ว

      tried character ai?

  • @mauricioweber8879
    @mauricioweber8879 ปีที่แล้ว +9

    Could it be possible that all those questions were made by an AI faking voices to seem real?

    • @arifulislamleeton
      @arifulislamleeton ปีที่แล้ว +5

      Thank you so much

    • @User_yhvz
      @User_yhvz ปีที่แล้ว +3

      The AI bot recognizes your kindness ⬆️😂

    • @yumi8106
      @yumi8106 ปีที่แล้ว

      😂

  • @toofy7253
    @toofy7253 ปีที่แล้ว

    Keep it going please!

  • @arifulislamleeton
    @arifulislamleeton ปีที่แล้ว +2

    Hi I'm Ariful Islam leeton im software engineer and software developer and website development and Artificial intelligence and intelligence community United States

  • @Time_Chance
    @Time_Chance ปีที่แล้ว +1

    K Hao, please stop “uptalking”!

  • @auro1986
    @auro1986 ปีที่แล้ว +1

    wsj, make one artificial intelligence program ask questions to another artificial intelligence program

  • @林為正
    @林為正 ปีที่แล้ว

    Without AI.

  • @mrmakeshft
    @mrmakeshft ปีที่แล้ว

    What I wonder is if the chatbot is going to develop into a cat and mouse game or is a chatbot going to be like a literary assistant

  • @林為正
    @林為正 ปีที่แล้ว

    To know if students are copying Chat GPT, just ask them to write something IN CLASS.

  • @mrmakeshft
    @mrmakeshft ปีที่แล้ว

    So I view the copyright thing as a multiple sided issue. And technology copies is an issue that is Always going to be subject to time. An artist that physically picks up a pen or etc. Is no different than digital so that’s a fight that is timely

  • @tonicalloway7227
    @tonicalloway7227 ปีที่แล้ว +2

    I wonder if chat bot can fool the Allen Turing test..are you a robot?..

    • @3dus
      @3dus ปีที่แล้ว +5

      GPT 4 has passed several turing tests. The issue now is that they dont know how to test it anymore.

  • @jonathanlatouche7013
    @jonathanlatouche7013 ปีที่แล้ว

    Software and hardware

  • @SaurabhOKumar
    @SaurabhOKumar ปีที่แล้ว

    Can ChatGTP be programmed on reverse psychological parameters, in a way, to read it's own human level proficient logic.. in a pursuit to understand the process of its own dynamic approach as an end result, rather than, the actual results... in a way, to logically understand logical variances and deviations it is taking to bring in the automata instincts in itself that it registers to evolve as a better decision maker and grasper for the problems it is continuously solving over time and hence, becoming better than before? In other words, can a ChatGTP stand against two other ChatGTP facing each other wherein the first gives the result for any given query, the second targets for the logic of results efficiency of the first one while, the third targets for logic of the process and procedures taken to bring that result given by the first? In a way creating a loop which will help human beings profile the growth and evolutionary steps taken by ChatGPT to solve human queries and thereby, making us understand the limitations of the human thinking process that makes "the man" pop-up those questions and rely heavily on these tools.

  • @grerovambrozoyuz9426
    @grerovambrozoyuz9426 ปีที่แล้ว

    If Ai have hallucination meaning he is Concious..

  • @jonathanlatouche7013
    @jonathanlatouche7013 ปีที่แล้ว

    18:38

  • @jonathanlatouche7013
    @jonathanlatouche7013 ปีที่แล้ว

    Authentiqie

  • @EveningDew001
    @EveningDew001 ปีที่แล้ว +1

    My Question is.. how do I get rid off your reporter sitting under my phone? I have a metro PC phone. And Mr. Zulfiqar Ali BHUTTO grandson Michael ZULFIQER and his DIRECT TV via Real News Network bots hear all my convo.

  • @cosmicjerry8570
    @cosmicjerry8570 ปีที่แล้ว

    🤖🚬 ☕☕☕

  • @stein1385
    @stein1385 ปีที่แล้ว +1

    this is better off read in text than listened to... but whatever wsj wants to platform for the sheep they go with oh well thats sad

  • @being1me
    @being1me ปีที่แล้ว +6

    Train it to “follow our values”- well that’s dangerous.
    And companionship/relationship with AI? Whew this is really something… 🫠