Mercilessly pricking the bubbles of AI, Big Data, machine learning


At the end of the day AI is just , an educated guess , made by a machine for a human .


It’s almost ironic when these things actually do something after the hype they receive. There have been plenty of useful expert system applications. There’s just not much magical about them, and they were inflated to a hundred times their proper proportion by the hype machines of the 80s…


This is in fact the traditional definition of AI.


Inflated fifteen to one hundred times, and promised at least five fold earlier than they actually arrived. If you hear that some technology will transform the world in 5 years, plan on waiting 25, like the PC.


I think the current enthusiasm for AI is largely driven by the near-term availability of hardware capable of emulating brain-sized complexity. For the first time, researchers will be able to casually train multiple adaptive behaviors that require a few thousand neurons each, and assemble them into a working whole that has many learned behaviors across disparate domains, and be able to leverage learned behavior from one domain onto new ones.

We finally have enough snow, and the slope seems about right, perhaps all we need is a big enough snowball with enough mass to take off rolling down the hill of experience, adding newly trained submodules as new experiences are encountered. Just gotta build that first big snowball and get it rolling.


Beats being the “other” Michael Bolton. Or Tiger Woods for that matter.


The author is decrying the sensationalism of the article. This should be read also to understand what he meant:


This topic was automatically closed after 5 days. New replies are no longer allowed.