Artificial Artificial Artificial Intelligence

Artificial Artificial Artificial Intelligence

Artificial Artificial Artificial Intelligence

 

Everything has been thought of before, said my mad scientist Polish friend once.  He might be right and maybe what I’m going to say has been thought of before.  

Take Mechanical Turk.  The paradigm is Human Intelligence Tasks (“HITS”), tasks that are menial (or mostly menial) and yet it is deemed only humans can complete them.  

The operative word is *human*.  My thesis below is:  let’s chuck out *human* and replace it with *human* or *machine*.  We'll call it XIT.

By the way, on Mechanical Turk, some UBS colleagues of mine and I thought of creating such a concept in 1998; we called it the Human Search Engine.  I as well as the others backed out but there was one person (one of those high IQ ubermensch fluid intellect types) who had lots of ideas and he was gung-ho. It was ahead of its time but I took a very high-paying day job instead.  

In any event, for those of you who have  reviewed some of the requested HITS from the perspective of someone who wants to request such HIT work, doesn’t it strike you as odd that there seems to be a prejudice against machines?  Why not let a machine, or someone’s machine, do the HIT work.  If someone wants to pay 1000s of humans to classify images as being scary or nice or whatever, and some other person already possesses a highly trained algorithm to classify such images with a similar degree of accuracy afforded by human classification, why should there be a bias against the machine?  

From my perspective, if machines were allowed, then this could give rise to transparency in the (lower-bound) valuation of proprietary algorithms.  If someone is willing to pay 10,000 dollars to do some work that a machine in someone’s possession could do almost instantaneously, then doesn’t that put a lower bound on the value of that machine’s algo? I think so.   In a sort of Cyber-punk,  proto-apocalyptic sense, the concept could give rise to an “exchange” for the pricing and trading of algorithms themselves, based on valuations by requests from XIT frameworks.

The other aspect of this is, unlike the world of hedge funds, where you could have a Fund of Fund of Fund of Fund of hedge funds, getting paid to invest in Fund of Fund of Fund of hedge funds  getting paid to invest in Fund of Fund of hedge Funds, getting paid to invest in hedge funds, in this set up it is clearly idempotent.   That is to say, it seems to me that if there were a Mechanical Mechanical Turk that farmed out algos to a Mechanical Turk, there is no added-value in doing so (well, that’s true in Alternative Investments, but entities still exist to facilitate intermediaries).   One layer of a “Mechanical Turk” mechanism that allows for XIT, X standing for either Human or Machine should suffice.

Anyway, this must have been thought of before….happy to hear your feedback.

要查看或添加评论,请登录

Marc Nunes的更多文章

  • A Sentimental Invitation: Part Clinton

    A Sentimental Invitation: Part Clinton

    So, I've been staring at my API and HTLM/javascript output of sentiments on Trump, Clinton, Sanders and Brexit. If you…

    4 条评论
  • A Sentimental Invitation: Part Brexit

    A Sentimental Invitation: Part Brexit

    It's been a while since I wrote on this series. I won't talk about applying category theory and explaining inaccuracy…

    10 条评论
  • Getting the news before the news

    Getting the news before the news

    Firstly, and importantly, I hope there were no serious injuries in today's earthquake in California. Earthquakes are…

    3 条评论
  • Being on Time

    Being on Time

    Call me old skool. This will be short.

    8 条评论
  • Ideas

    Ideas

    This will be terse. I used to ponder for some time on the following simple question: how many ideas have been pondered…

    1 条评论
  • Subscribe, Mine, Publish

    Subscribe, Mine, Publish

    Mining Equipment Or Mining? A core grade school lesson you learn as a young child growing up in California is that it’s…

    1 条评论
  • Happy New Year from Amsterdam

    Happy New Year from Amsterdam

    4 条评论
  • Systemic Risk

    Systemic Risk

    Keystone Cops As they say, people have short memories. People still talk about the 2008 Financial Crisis but it seems…

    12 条评论
  • A Sentimental Invitation Part III-b

    A Sentimental Invitation Part III-b

    Recap To Recap, Part I, I was the evangelist, Part II, I constructed a viable simple Na?ve Bayesian Classifer, Part…

    7 条评论
  • A Sentimental Invitation Part III-a

    A Sentimental Invitation Part III-a

    A recap thus far. In Part I, wore the Evangelist Hat.

    1 条评论

社区洞察

其他会员也浏览了