Close Menu
    Zukunftsforschung

    Die 99-prozentige Erfolgsquote des GEN-1-Modells des Roboterunternehmens Generalist zeigt uns, dass humanoide Roboter schneller Fortschritte machen, als die meisten Menschen erwarten.


    "Mit GEN-1 haben die physischen Modelle jedoch laut Generalist einen GPT-3-ähnlichen Wendepunkt erreicht, an dem einige Aufgaben beginnen, „das Leistungsniveau zu überschreiten, das für den Einsatz in wirtschaftlich sinnvollen Umgebungen erforderlich ist“.

    Ich denke, dass humanoide Roboter einer der Schläfer-Technologietrends sind, die die meisten Menschen unterschätzen. Für die meisten ungelernten und angelernten Arbeiten benötigen sie weder AGI noch „perfekte“ KI. Bei ausreichender Entwicklung und Schulung werden die heutigen KI-Modelle wahrscheinlich in Ordnung sein. Hier ist ein weiteres Zeichen dafür, dass diese Hypothese wahr sein könnte.

    Wie schnell werden sie dort ankommen? Bei den derzeitigen Entwicklungsraten scheint das Jahr 2030 eine vernünftige Schätzung für Allzweck-Humanoide zu sein, die sich leicht für die meisten ungelernten/angelernten Arbeiten ausbilden lassen. Gerade dann, wenn die meisten Fahrerjobs durch Robo-Taxis verschwinden werden. Niemand scheint auf diese Zukunft vorbereitet zu sein, die schnell auf uns zukommt.

    Vom Falten von Kartons bis zum Reparieren von Staubsaugern erreicht das GEN-1-Robotikmodell eine Zuverlässigkeit von 99 %: Das neue Modell kann auf Störungen reagieren und Bewegungen erkennen, für die es nicht trainiert wurde.

    The 99% success rate of the Robotic company Generalist's GEN-1 model shows us that humanoid robots are progressing faster than most people expect.
    byu/lughnasadh inFuturology

    View 15 Comments

    15 Kommentare

    1. gunlamar on

      Maaan now this degree in low skill labour is useless!?! Shoulda went to school for robotics or some shit

    2. daemon_panda on

      I get laughed at when I tell people that most jobs will be replaceable in 7 years. I now wonder if that is not soon enough

    3. Apprehensive-Box-8 on

      Is the humanoid robot in the room with us?

      This is about an AI model for robots to do stationary tasks, not allowing a humanoid robot to move freely in a packed environment.

      It is a very big step in terms of robotics, yes. It has absolutely nothing to do with humanoid robots.

    4. Any-Individual5262 on

      I mean at factories or some gig work like delivery this robots can take over.

      Then it will start to move into skilled blue colour territory like carpenter or plumber.

      Eventually I believe it will move into highly skilled territory like craftsman.

    5. zeusomally on

      It is stunning how few politicians and economists are talking about this.  It should be front page news every single day.  And we also need to understand that most skilled work will also be replaced by humanoid robots faster than most people realize.   

    6. notmyrealnameatleast on

      It’s the perfect time to become communist or socialist as a country, so the people can share the bounty of free labour and basically free energy from solar and other green sources.

      If all robot and automated work becomes taxable like if a person did the job, then the governments can build a sort of welfare state where people who don’t have a job can get free education, healthcare, transport, and money to live while they’re sick or out of work etc.

      We can make a beautiful future, we can do it, we just gotta speak our mind and support the good things instead of losing hope and focusing on the bad too much!

      Let’s do it! Let’s make a utopia!

    7. botsmy on

      i ran a small warehouse pilot last year with two of the older-gen robots doing box sorting, and our error rate dropped from 6.3% to 1.8% after just three months of model updates. it wasn’t flashy, but that drop cut our rework labor by nearly half, which i didn’t expect so fast. makes me think these things don’t need to be perfect to start saving real money.

    8. m3kw on

      99% success rate in a controlled sandboxed environment designed to prevent other variables from causing golden path failure

    9. gminx on

      99% success rate on cherry-picked stationary tasks in a controlled setting is not the same number once you add navigation, clutter, and a human walking through the workspace.

    10. Euphoric_Gas9879 on

      Robots have been doing extremely skilled jobs for about 40 years. Ever seen a Toyota factory?

    11. MyTnotE on

      I believe we have about 14 years left before the average person is completely replaceable.

    12. Thewrongthinker on

      I told my nephew to get trained as Robot fixer. The job of the future. At least they become single used robots or create a robot that fix the broken robots but who fill fix the robot that fixed the broken robots and so on.

    13. smack54az on

      What happens the other 1% of the time? 99% sounds good until think about how many of these basic action a person does in an hour or a day. I read tge article and it says the model recovered from mistakes, but what if those mistakes compound? Also I’m uncomfortable with a machine in my home that requires a constant connection to it’s controlling model data center constantly feeding my life into the training algorithm. It’s a security nightmare.

    14. VoidOmatic on

      So it’s good enough that it can accidentally kill people? Nothing bad is going to happen with this.

    15. Fredasa on

      > No one seems prepared for this future rapidly bearing down upon us.

      I mean, we _are_ preparing, right? By reducing the number of babies born?

    Leave A Reply