Never forget to read the alt-text joke on an XCKD comic. The punchline on this one is based in some truth.
> In the 60s, Marvin Minsky assigned a couple of undergrads to spend the summer programming a computer to use a camera to identify objects in a scene. He figured they'd have the problem solved by the end of the summer. Half a century later, we're still working on it.
What's really incredible to me is how right the prediction was, but also that it was real. In 2014, classifying images as "containing a bird" was a nigh-on-impossible task. Not an impossible one, and image classification was already in production use in limited forms with mapping agencies and the like, but beyond anyone's capability at the time. In 2017, Not Hotdog was a novelty app - Image classification was real, but limited, and didn't have a great reputation yet. By 2019, papers[1] were being written on image classification as a service and where their pitfalls were, but the idea was solid and sound; Today in 2022, it's something you'd have to research and test before buying for your startup but not a hard product to find.
> In the 60s, Marvin Minsky assigned a couple of undergrads to spend the summer programming a computer to use a camera to identify objects in a scene. He figured they'd have the problem solved by the end of the summer. Half a century later, we're still working on it.
https://dspace.mit.edu/bitstream/handle/1721.1/6125/AIM-100....