1. Welcome! Please take a few seconds to create your free account to post threads, make some friends, remove a few ads while surfing and much more. ClutchFans has been bringing fans together to talk Houston Sports since 1996. Join us!

[DOOMED] Scientists Really Don't Understand How AI Works

Discussion in 'BBS Hangout' started by Xerobull, Mar 6, 2024.

  1. Xerobull

    Xerobull You son of a b!tch! I'm in!

    Joined:
    Jun 18, 2003
    Messages:
    33,405
    Likes Received:
    30,972
    SCIENTISTS HAVE A DIRTY SECRET: NOBODY KNOWS HOW AI ACTUALLY WORKS

    [​IMG]

    "OUR THEORETICAL ANALYSIS IS SO FAR OFF WHAT THESE MODELS CAN DO."
    GETTY / FUTURISM

    Mystery Machine

    The AI industry exploded in late 2022, following the wild success of OpenAI's ChatGPT release in November of that year. Generative AI tools — from chatbots to remarkably lifelike music and voice-generators to image and video creators, and more — continue to dazzle the public, while AI and machine learning advancements continue to find applications in fields like healthcare and drug discovery.

    Just one problem: not even the folks creating all this AI fully understand how it really works.

    "Obviously, we're not completely ignorant," University of California, San Diego computer scientist Mikhail Belkin told MIT Technology Review. "But our theoretical analysis is so far off what these models can do."

    Indeed, as MIT Tech explains, many AI models are notoriously black boxes, which in short means that while an algorithm might produce a useful output, it's unclear to researchers how it actually got there. This has been the case for years, with AI systems often defying statistics-based theoretical models. Regardless, the AI industry is careening ahead, fueled by billions of investment dollars and a hefty share of near-fanatical belief. (And, of course, the C-Suite vision of eliminating vast swaths of the workforce.)

    In other words, AI is already everywhere. But as it's increasingly integrated into human life, the scientists building the tech are still trying to fully understand how it learns and functions.

    Spaghetti, Meet Wall
    Some experts chalk the lack of understanding up to the burgeoning nature of the field, arguing that AI's nascency means that sometimes researchers will have to work backward from experimental results and outputs.

    "These are exciting times," Boaz Barak, a Harvard University computer scientist, told MIT Tech. "Many people in the field often compare it to physics at the beginning of the 20th century."

    "We have a lot of experimental results that we don't completely understand," Borak continued, "and often when you do an experiment it surprises you."

    No Guarantees
    If the industry's still-minimal degree of understanding seems at all cavalier, though, it's because it is.

    To be sure, experimentation and uncertainty are natural, if not inherent, to the scientific process. But AI models are no longer restricted to metaphorical Silicon Valley test tubes; the AI industry is a lucrative financial behemoth, and like with other technologies, the familiar "move fast and break things" approach that many in the industry have moved with could well present challenges down the road.

    After all, as Belkin told MIT Tech, guarantees are important. And here? They don't exist quite yet.

    "I'm very interested in guarantees," said Belkin, according to the report. "If you can do amazing things but you can't really control it, then it's not so amazing."

    "What good is a car that can drive 300 miles per hour," he added, "if it has a shaky steering wheel?"
     
  2. Invisible Fan

    Invisible Fan Contributing Member

    Joined:
    Dec 5, 2001
    Messages:
    43,372
    Likes Received:
    25,378
    It really is a Frankenstein. The reason why Altman wanted to raise "7 Trillion" is so he could get the chips and compute to make whatever's happening get smarter.

    Mmmmm Spaghetti Meat Wall.
     
    clos4life and Xerobull like this.
  3. ElPigto

    ElPigto Member
    Supporting Member

    Joined:
    Sep 21, 2006
    Messages:
    14,777
    Likes Received:
    23,344
  4. MadMax

    MadMax Contributing Member

    Joined:
    Sep 19, 1999
    Messages:
    73,559
    Likes Received:
    19,851
    “TO THE IDIOTMOBILE!!”
     
    BamBam likes this.
  5. Xerobull

    Xerobull You son of a b!tch! I'm in!

    Joined:
    Jun 18, 2003
    Messages:
    33,405
    Likes Received:
    30,972

    Outstanding. Two Simpsons references in a row.
     
    ElPigto likes this.
  6. boomboom

    boomboom I GOT '99 PROBLEMS
    Supporting Member

    Joined:
    Sep 29, 1999
    Messages:
    12,261
    Likes Received:
    8,630
    AI will always have one intrinsic flaw...no matter how many iterations of AI generating new AI, the original Creator will always be a human...and those included flaws will never be wiped from it.
     
  7. Rocket River

    Rocket River Member

    Joined:
    Oct 5, 1999
    Messages:
    61,640
    Likes Received:
    29,054
    Not unlike the human brain
    We put enough information in it . .. that it becomes unpredictable
    Right now its pure digital information
    Images, sounds and works/letters
    Add smell, taste and feel . . . then you get like a replica of human results IMO

    Within 10 maybe 20 yrs. . . . AI Psychologist will be a thing

    Rocket River
     
  8. clos4life

    clos4life Member

    Joined:
    Dec 20, 2007
    Messages:
    10,871
    Likes Received:
    11,809
    Know your enemy. We're failing. Doomed.
     
    TimDuncanDonaut and boomboom like this.
  9. droxford

    droxford Member

    Joined:
    Oct 26, 2001
    Messages:
    10,105
    Likes Received:
    1,559
    I work with AI engines.

    The business worlds is all gushing about how great it is. People think it’s such a great tool.

    But if they really understood AI, they’d be afraid of it. Right now, it’s a tool that can help us. Soon, it will replace us. A LOT of people are gonna lose their jobs to AI.
     
  10. Invisible Fan

    Invisible Fan Contributing Member

    Joined:
    Dec 5, 2001
    Messages:
    43,372
    Likes Received:
    25,378
    Better hail our robot overlords now, or said robot overlords will hail you for an Uber ride.

    BTW, the layoffs happening in tech hasn't affected Google or Big Tech's overall head count as much as the headlines suggest.

    They are totally restaffing with AI specialists in mind...
     
  11. ThatBoyNick

    ThatBoyNick Member

    Joined:
    Dec 8, 2011
    Messages:
    28,445
    Likes Received:
    43,639
  12. ThatBoyNick

    ThatBoyNick Member

    Joined:
    Dec 8, 2011
    Messages:
    28,445
    Likes Received:
    43,639
    This is supposedly the future many dreamed of though, with robots doing most of the grunt work while humans have more time for living with leisure, family, arts, what name you.

    The question remains in a world with rapid technological/industrial automation, in a world with rapid wealth accumulation by an incredibly small number of people, how are we going to avoid the obvious (seemingly) looming dystopian trap of modern feudalism with the new era kings being the owners of the machines

    For the dream of automation finally freeing humanity from menial work to live, there has to be a system in place to distribute the resources/value that the machines produce to all. That, or everybody but eventually a handful can eat **** I guess.
     
    clos4life likes this.
  13. Invisible Fan

    Invisible Fan Contributing Member

    Joined:
    Dec 5, 2001
    Messages:
    43,372
    Likes Received:
    25,378
    There's been ideas to tax big tech to fund Andrew Yang's UBI.

    I think something that involves the digital equivalent of Georgist land taxes would be better than the government smashing and grabbing their own private AI or being captured by those companies through insider trading (or worse).

    My idea would be that you'd tax AI companies as if it were developed land based on the total compute it has. Privately run LLM models for small businesses or non-profits wouldn't need to be taxed if they don't surpass that threshold.

    By counting aggregate compute over something like GPT-4's marketshare, you would then avoid the loophole of a company breaking up the AI into smaller service-specific AIs.

    It'd be a form of progressive taxation without killing innovation and private industry.

    Henry George considered land a scarce and public good with land speculators (Google: "shadow inventory") as a scourge that promotes economic bubbles. Because AI hoovers our data like no tomorrow without proper recompense, it consumes us as much as we consume it.

    There's likely holes in this idea much like how there were holes in Georgism during boom times, but any public debate over it would be new debate.

    https://www.econlib.org/library/Enc/bios/George.html
     
    #13 Invisible Fan, Mar 7, 2024
    Last edited: Mar 7, 2024
    rocketsjudoka likes this.
  14. PhiSlammaJamma

    Joined:
    Aug 29, 1999
    Messages:
    28,760
    Likes Received:
    7,043
    Our first great triumph was tricking women into wanting to do work. Done.
    Our second great triumph is getting AI to do work. Done.

    AI will break through the glass ceiling, and we'll laugh as they do it. I honestly don't care how it does the work. Just do it for me. Go hog wild.

    Anything that is stupid enough to work can do it.

    Cats don't do shiot. They are the secret overlords.
     
  15. Xerobull

    Xerobull You son of a b!tch! I'm in!

    Joined:
    Jun 18, 2003
    Messages:
    33,405
    Likes Received:
    30,972
  16. Rocket River

    Rocket River Member

    Joined:
    Oct 5, 1999
    Messages:
    61,640
    Likes Received:
    29,054
    That's my point. When people says they will replace us . . . .what does that look like?

    I think what will happen is that we will reach a point where human interaction may not even be a thing
    When your every whim is satisfied by AI/Robots do you need other people?
    What will become of humanity?
    When you can sit on your *ss all day . .eat whatever you like. . . f*** Sex bots that adore you and look like whoever
    (until these bots get old . .. . then you buy a new one and the old one now needs alimo . ..er. . . . I keed I Keed , LOL )

    But seriously . . . when every Physical NEED is met . . . .. What does a human do?

    Rocket River
     
  17. droxford

    droxford Member

    Joined:
    Oct 26, 2001
    Messages:
    10,105
    Likes Received:
    1,559
    In the more immediate future... here's what that will look like...
    Ya know how, in the workplace, you don't actually build a physical server any more - you spin up a new virtual server, right?
    With AI, you will spin up a virtual worker, instead of hiring someone.
    Need two more marketing team members? You'll spin up two virtual workers and there ya go.

    Generally, if your job doesn't require you to do anything physical, you're going to lose your job to AI.
    Marketing
    Sales
    Support
    Artists of all kinds: actors, writers, singers
    The entire movie industry will change to CGI. It will be MUCH cheaper, faster, easier to create a movie with virtual actors acting out a AI-written plot with computer-generated soundtrack. Won't have to pay actors, directors, sound staff, costume, props, makeup, lighting.. won't have to go on location or use cameras... the quality won't be as good (at first), but because of the immense cost savings, the traditional method of making films will become much too cost-prohibitive.
    At first, lawyers will use AI as a helpful tool, but then AI will replace lawyers (maybe not all of 'em, but many of 'em).

    ....and the physical jobs that take skill, like welding, woodwork, etc. will all be taken up by low wage illegal aliens.
     
    Rocket River likes this.
  18. droxford

    droxford Member

    Joined:
    Oct 26, 2001
    Messages:
    10,105
    Likes Received:
    1,559
    ...and it doesn't get better for human relationships. Why put all the work into a long-term relationship with a human being when you can have pure long-term satisfaction from an AI spouse? With an AI spouse, you can have anything you want, any time, without compromise.... (except, of course, childbirth).
     
  19. Rocket River

    Rocket River Member

    Joined:
    Oct 5, 1999
    Messages:
    61,640
    Likes Received:
    29,054
    I was looking at SORA
    We are less than a decade away
    from being able to scan in Lord of the Rings and it will churn out a whole accurate movie

    Rocket River
     
    droxford likes this.
  20. droxford

    droxford Member

    Joined:
    Oct 26, 2001
    Messages:
    10,105
    Likes Received:
    1,559
    I had not heard of SORA.

    Wow. Yup - that's what I'm talking about. This is just the beginning - AI is in it's infant stage of growth.
    Right now, we laugh about how the images that AI makes are garbled and we gush about how powerful of a helping tool it is.
    In a few years, it's gonna be a lot different.
     
    Rocket River likes this.

Share This Page

  • About ClutchFans

    Since 1996, ClutchFans has been loud and proud covering the Houston Rockets, helping set an industry standard for team fan sites. The forums have been a home for Houston sports fans as well as basketball fanatics around the globe.

  • Support ClutchFans!

    If you find that ClutchFans is a valuable resource for you, please consider becoming a Supporting Member. Supporting Members can upload photos and attachments directly to their posts, customize their user title and more. Gold Supporters see zero ads!


    Upgrade Now