I'm frequently amazed at what non-technical people believe might be possible with things like software and computers. Opinions are often wildly optimistic for things that are near impossible (currently) and pessimistic for things that are trivial, there's almost an inverse correlation between reality and imagination. I blame Hollywood, the constant stream of teenagers breaking into military computers by "guessing" the single password on their second attempt is somewhat misleading, however sometimes what people think should be easy (because it is for a Human) turns out to be really difficult for reasons that are opaque to them.
This xkcd cartoon sums things up nicely,
But, here's the rub, this cartoon was created 6-7 years ago and since then advances in AI have progressed to a point where it would take just a few minutes to tackle this specific problem, there are many reasons for this, mainly huge investments in time and energy in creating libraries of training data (involving pictures of birds) that have been built up over the years (it wasn't a single eureka moment!) Such is the nature of progress and one of the main reasons people struggle to keep up with the reality of what's computationally possible and what's not. We can calculate the gravitational force that the Eifel Tower imparts on a baseball in Boston but we still can't model the economy or the weather or even which team will win the World Cup particularly well, some things are just too complex and appreciating complexity requires an clearly requires an investment in understanding.
No comments:
Post a Comment