Saturday, February 19, 2022

After 30 years, they're still doing it!

In Today's New York Times: an example of text describing an image.
Guess what image they chose as their front-page example? 


I would have thought that people writing about computers now would have moved on, but the front page of the New York Times this morning shows otherwise.

"Use Mona Lisa in your computer ad" was the title of a little essay I wrote in the mid-1990s. Mona Lisa often appeared as an example in computer ads and in articles on how to use the emerging computer graphics software at the time. I made the point that Mona Lisa had very high recognition among computer programmers, that as a woman she had appeal to these (mostly) men, and that she was non threatening to these (mostly) nerds. In fact, a Mona Lisa rendered in numbers was one of the very first digitized images ever made – in 1965 – created to be printed out on a very early printer.

Mona Lisa's Eyes, from the Mona Lisa that was digitized in 1965. (source)

Byte magazine, 1987 cover.

I guess it's not surprising that even now, Mona Lisa is the first example you would see in an article on artificial intelligence applied to image recognition. The only surprising thing is that after more than fifty years, AI can't actually recognize the Mona Lisa and give her name!! Well, some of the AI recognition programs are doing a little better with food images:

Another example of machine-generated descriptions from the article.

Those nerds will never change. See the NYT article "The Hidden Image."

Blog post © 2022 mae sander.

8 comments:

  1. Hello

    Mona Lisa has to be the most recognizable painting/person on the internet.
    LOL, I do not think I would mistake a plate of pancakes for a slice of pizza.
    Have a great day!

    ReplyDelete
  2. Apparently, AI has a way to go to be truly intelligent. Even facial recognition software is better than a strawberry mistaken for a pizza. I found humor in this.

    ReplyDelete
  3. I'm still trying to figure out what disability rights activists has to do with hidden image descriptions challenges?

    ReplyDelete
  4. The first paragraph of the article explains why photo descriptions help people with disabilities:
    "The text boxes above are examples of alt text, which people who are blind or have low vision often rely on when navigating the web. When it’s available, the text can be detected and read aloud or translated into Braille through screen readers, assistive technology that can be accessed in the form of software programs, apps or even browser extensions. For these users, alt text is essential to the online experience."

    ReplyDelete
  5. OK -- where's the strawberry?

    ReplyDelete
  6. I saw this NYTimes article, too. The Mona Lisa certainly seems like one of the most recognizable images, at least to the "western/European-influenced" nations

    ReplyDelete
  7. Thank you for the explanation in your comment answer above. I too did not make the connection to disability rights. I keep thinking I should fork over some more money to the NYT because I believe they do good work -- but I already have so much to read. (I have a "baby" subscription to the Times, for cooking and puzzles and that lets me read some, but not all, of the "real" newspaper. ) This was a very interesting read anyway. I remember when Google ran a kind of game asking you to help by identifying random photos way back when they were first putting their "search image" into effect. I think you got to move on if your answer matched the most other answers.

    ReplyDelete
  8. The pizza guess is intriguing lol

    ReplyDelete

Thank you for commenting. Please include a link to your current blog so that I can read your blog and share more of what you are thinking. Your google-blog-ID may not link to a blog hosted at another site, so please let me know who you REALLY are!