Tuesday, August 9, 2016

AI is everywhere. Just ask anyone.

This was on the John Oliver show this week. As he points out, this is pure nonsense:







Of course, we could just note that the speaker is just ignorant about what AI means and is really talking about machine learning. Here is an excerpt from the Wikipedia entry on machine learning:

Machine learning is closely related to (and often overlaps with) computational statistics; a discipline which also focuses in prediction-making through the use of computers. It has strong ties to mathematical optimization, matrix theory, linear algebra, and copulas, which delivers methods, theory and application domains to the field. Machine learning is employed in a range of computing tasks where designing and programming explicit algorithms is unfeasible. Example applications include spam filtering, optical character recognition (OCR),[5] search engines and computer vision. Machine learning is sometimes conflated with data mining,[6] where the latter sub-field focuses more on exploratory data analysis and is known as unsupervised learning.[4]:vii[7]


But can statistical approaches to text processing do what the speaker suggests? Yes. It is called search and journalists already do this. Could a computer automatically do search? I am not even sure what that means. A computer can’t do anything that it hasn’t been told how to do and for which clear parameters haven’t been specified. Getting the computer to “automatically” find photography doesn’t sound so difficult. And it is quite easy if you are looking for a picture of John Oliver for example. Or is it?

Here are some pictures that appear on Google images search for “John Oliver.” 











How would the machine know which of these pictures are appropriate for a newspaper to use? It wouldn’t.

As AI heads for its inevitable winter due to over promising, I am hoping for Global Warming (of investors.)






No comments:

Post a Comment