Is AI going to change the world?

by Reasonfirst 54 Replies latest jw friends

  • Terry
    Terry

    The "limitation" of A.I. is IMPOSED and not really an artifact or bug.
    Put your dog on a leash fastened to a tree and then command him to fetch
    the stick you toss across the street. (That sort of thing.) Coming up short
    is built in.
    Hallucination is giving an answer when there is no data and suddenly imagination
    kicks in. If you have parented a three year old you'll recognize this tendency.
    Call it what you like (lies, hallucinations, etc) but it is temporary.
    Artificial General Intelligence is the most useful 'tool' humans have ever produced.
    Those of us who try to stay in the loop as miracles appear will not fear it but wisely
    seek to adopt it, utilize it and make life so much less burdensome.
    We all have friends who just won't budge from what they already know; who resist,
    chide, criticize, flail and fulminate against every new technology. They will prosper
    least, I fear.

  • TD
    TD

    Judgement, as in understanding your information may be biased or inaccurate in any number of ways manifests itself in a human physician's ability to discern that a patient's symptoms may be psychosomatic or misreported, or a human engineer's ability to determine an Architect may be mistaken on anything from materials to the application of mathematics, or a human judge's ability to interpret the intricacies of existing case law.

    AI lacks judgement and there is not a consensus of opinion even within the field on how this hurdle may be overcome.

    I'm not saying it can't be overcome, but I am saying that quantum leaps can't be predicted or scheduled.

  • joey jojo
    joey jojo

    For all its faults, A.I. is still pretty amazing right now. Businesses of all types are being encouraged to adopt A.I. into their models, or get left behind.

    The danger might lurk in the future as A.I. becomes more advanced and bypasses human decision making. If various A.I. systems were talking to each other it might not be sentient or intelligent like Skynet, just that the potential for catastrophe might have increased.

  • TonusOH
    TonusOH

    I think the more logical concern is that of poorly-configured or out-of-date models. We know that buggy software can cause all kinds of problems, and we know that some of them are obvious (crashes, graphical glitches) and others can be very subtle yet dangerous (incorrect output that can have catastrophic long-term effects).

    Slim mentioned that AI is already doing a good job with human hands when it generates art. This sort of relatively obvious change will be a good way to see who is using outdated AI to generate art. In time, we may be able to tell which version of AI they are using by the quality of the work. In other fields, using outdated or poorly-designed AI can create problems that we don't see right away, but that can be significant nonetheless.

    As with just about every human discovery and invention, we are likely to learn a few lessons the hard way.

  • Reasonfirst
    Reasonfirst

    An aspect of AI that I had not considered, mainly because it's the first mention I've seen of this issue.

    This is an extract from an article in the HK based South China Morning Post:

    "China is joining forces with Asean and the United Nations to tackle the scam networks that have dug in across Southeast Asia, the UN Office on Drugs and Crime said on Tuesday, as it warned that the region’s criminal gangs were becoming more powerful, networked and were adopting AI to widen their pool of victims."

  • Earnest
    Earnest

    Unfortunately AI is also being used in CSA Media as shown by this Internet Watch Foundation report.

  • TonusOH
    TonusOH

    Criminals will usually be the first to exploit new technologies. And with so much cybercrime happening, I suspect that they might even drive the development of AI. Probably not a good thing...

  • Reasonfirst
    Reasonfirst

    Seems to be an improvement in processing power:

    Shared from The Conversation. The authors are Zdenka Kuncic is Professor of Physics, University of Sydney and Ruomin Zhu is a PhD student, University of Sydney

    https://theconversation.com/we-built-a-brain-from-tiny-silver-wires-it-learns-in-real-time-more-efficiently-than-computer-based-ai-216730

  • Simon
    Simon

    Ai is great for some things, but it's a long way off even basic intelligence.

    Basically, it's still thick as pig-shit.

    Unfortunately, so are most people, so they are easily fooled into thinking that some pattern matching is "clever" without realizing that there is zero reasoning happening.

  • slimboyfat
    slimboyfat

    Yes, zero reasoning is happening, but that doesn’t matter. What matters is outcomes. A computer can beat you at chess, solve protein folding that has eluded the best efforts of scientists for decades, and it can write more persuasive copy than any human can, and all the while zero reasoning is happening, only number crunching. Machines are not becoming conscious they are becoming capable, and their capability is increasing at an exponential rate. If it seems like they have a long way yet to go then remember that an exponential curve is already half way to a destination when only a tiny fraction of the ground has been covered. The story goes that it took scientist years to map just 1% of the human genome and despondent backers thought it would never be competed in their lifetimes. Somebody who knows how exponential progress work responded: “1% alread? Then you’re almost done!” And it was completed shortly after.

    GPT2 produced incoherent strings of words

    GPT4 is already passing the Turing test of output that humans find difficult to distinguish from a human being

    What will GPT5 and 6 and 7 be like?

    The world is going to change radically beyond recognition within a very short period of time.

    https://youtu.be/eD5GlCIS0sA?si=lchgY4xd7khmZ8P4

Share this

Google+
Pinterest
Reddit