This doesn't seem like a perspective offering anything new and possibly involving a few problems.
Jeff Hawkins' On Intelligence also emphasized the predictive qualities of intelligence but again, this was a new emphasis, not really a new idea but a new emphasis in that it's well known time is implicitly an element of nearly any cognitive process.
Moreover, the current deep learning driven AI renaissance is has been most effective in predicting patterns in quite static phenomena; image recognition and competitive video and board games (game involve change over short periods but their context remain identical over long periods). Similar success hasn't come in say, the stock market (so far as I know). So it seems increased prediction at the corporate level will come only as long as larger processes stay static (when you have cascade of change, you'll have cascades of failing predictors, which generate further change what could go wrong?).
> Similar success hasn't come in say, the stock market
Depends on your metric of success, I suppose.
I couldn't begin to imagine there aren't a bunch of ML tasks hammering away at something like a mutual fund where a conservative investment strategy could (probably, dunno?) be learned by an algorithm.
A seat of your pants day trading strategy learned by ML, probably not so much.
"There’s some number at which Amazon might think, “We are now sufficiently good at predicting what you want to buy. Why are we waiting for you to shop at all? We’ll just ship it.”
This is one of the most politically, economically, and psychologically tone deaf things I've read this year.
Hypotheticals don't have to be reasonable or realistic to serve their purpose.
The writer's purpose here was to show the difference between incremental and transformational business change, by stressing baked-in assumptions beyond their natural breakng point. In that sense they succeeded.
A better way to put it is: “We are now sufficiently good at predicting ... We’ll just ship it to the warehouse closest to you, so in case you buy it, you get faster shipping.” To me, that's value of AI.
Here's the author's pitch: The value of AI is in "cheaper prediction." So economic value capture occurs by looking for prediction problems in your business and unleashing AI on those. The author further prescribes that one may have to "recast" certain problems (e.g., self-driving cars) as prediction problems to make it all work.
But the stunner for me was the assertion that the value of the semiconductor was "cheaper arithmetic."
I find these over-generalizations neither accurate nor helpful.
Maybe mentions of all the crucial peripherals would have been relevant. The enormous storage capacity, the networks, high resolution displays and sophisticated interfaces.
The original quote was about semiconductors. Not computing systems. From which the author was drawing a thread to the profound impact a decrease in cost (or increase in throughput) of an input can have on the economy.
Taking an abstraction out of context and then griping about minutiae with "Well actually..." quips is intellectually dishonest.
If it's an inaccurate simplification, that's fair, but I thought it was valid for the domain the author was addressing.
Yeah, but nonetheless it will be shared on all public and corporate networks and influence business culture and decisions...
What world would we live in if nobody would have ever termed these technologies "artificial intelligence" but just stuck to "statistics" or maybe something equally boring as "advanced statistics" ...
12 comments
[ 3.1 ms ] story [ 22.0 ms ] threadJeff Hawkins' On Intelligence also emphasized the predictive qualities of intelligence but again, this was a new emphasis, not really a new idea but a new emphasis in that it's well known time is implicitly an element of nearly any cognitive process.
Moreover, the current deep learning driven AI renaissance is has been most effective in predicting patterns in quite static phenomena; image recognition and competitive video and board games (game involve change over short periods but their context remain identical over long periods). Similar success hasn't come in say, the stock market (so far as I know). So it seems increased prediction at the corporate level will come only as long as larger processes stay static (when you have cascade of change, you'll have cascades of failing predictors, which generate further change what could go wrong?).
Depends on your metric of success, I suppose.
I couldn't begin to imagine there aren't a bunch of ML tasks hammering away at something like a mutual fund where a conservative investment strategy could (probably, dunno?) be learned by an algorithm.
A seat of your pants day trading strategy learned by ML, probably not so much.
This is one of the most politically, economically, and psychologically tone deaf things I've read this year.
The writer's purpose here was to show the difference between incremental and transformational business change, by stressing baked-in assumptions beyond their natural breakng point. In that sense they succeeded.
Here's the author's pitch: The value of AI is in "cheaper prediction." So economic value capture occurs by looking for prediction problems in your business and unleashing AI on those. The author further prescribes that one may have to "recast" certain problems (e.g., self-driving cars) as prediction problems to make it all work.
But the stunner for me was the assertion that the value of the semiconductor was "cheaper arithmetic."
I find these over-generalizations neither accurate nor helpful.
In what way would you disagree with this?
For purposes of enterprise resource allocation (the article's domain), it seems an accurate simplification.
Taking an abstraction out of context and then griping about minutiae with "Well actually..." quips is intellectually dishonest.
If it's an inaccurate simplification, that's fair, but I thought it was valid for the domain the author was addressing.
Yeah, but nonetheless it will be shared on all public and corporate networks and influence business culture and decisions...
What world would we live in if nobody would have ever termed these technologies "artificial intelligence" but just stuck to "statistics" or maybe something equally boring as "advanced statistics" ...