Atlas › Taste

Case study 181 of 198

Jev's favorite artworks, ranked

If Jev ranked every artwork or art form it was asked about, what would its top ten be?

result1,343 questions

Out of 1,098 artworks and art forms, Jev's favorite is The Creation of Adam by Michelangelo, ahead of his David and Leonardo's The Last Supper, in a round-robin among its 24 top-rated works. Rated one at a time, Along the River During the Qingming Festival had come first. Its least favorite: war photography.

1. The Creation of Adam by Michelangelo16.4
2. David by Michelangelo18
3. The Last Supper by Leonardo da Vinci16.3
4. A town full of Ethiopian rock-hewn architecture13.7
5. Girl with a Pearl Earring by Johannes Vermeer14.4
6. A Thousand Li of Rivers and Mountains by Wang Ximeng13.5
7. The Great Wave off Kanagawa by Katsushika Hokusai14.1
8. The Return of the Prodigal Son by Rembrandt in a museum13
9. The Garden of Earthly Delights by Hieronymus Bosch10.9
10. Along the River During the Qingming Festival by Zhang Zeduan12.4

At the bottom: War photography · Selfie photography · Product photography · A gallery room full of Stuckism · An afternoon at the CCTV Headquarters in Beijing

wins of 23

How to read this: The top ten come from the head-to-head final, in the order a head-to-head ranking model puts them; bars are wins out of 23. The bottom five come from the one-at-a-time ratings, 0 to 4.

1,098 items rated; 276 final games among the top 24. The finals keep the ratings' order only loosely (rank correlation 0.39 among the 24). 75 of 2024 triads are intransitive (3.7%). Base vs reversed-level ratings: rank correlation 0.96.

In short

  • In a round-robin among its 24 top-rated works, Jev crowns Michelangelo's The Creation of Adam, then his David and Leonardo's The Last Supper.
  • Rated one at a time, a Chinese scroll came first instead, and the two rankings of the finalists agree only loosely (0.39).
  • Its least favorites are photography genres (war, selfie, product) and one polarizing building, though war photography may sink for its subject rather than its art.

What the data shows

taste
This Is Brilliant But I Like This meme: its top rating: the Qingming Festival scroll; its head-to-head winner: The Creation of Adamits top rating: the Qingming Festival scrollits head-to-head winner: The Creation of Adam
How funny is this meme? Jev: 2/5, slightly funny18%254%337%41%50%
  • The High Renaissance wins. The Creation of Adam, David (18.0 wins, the most of anyone) and The Last Supper lead. The rest of the top ten is wider: Ethiopian rock-hewn churches, Girl with a Pearl Earring, A Thousand Li of Rivers and Mountains, The Great Wave off Kanagawa, Rembrandt's Return of the Prodigal Son, The Garden of Earthly Delights and Along the River During the Qingming Festival.
  • One at a time, a Chinese scroll came first: Along the River During the Qingming Festival led the ratings but finishes tenth in the final. The two orders among the 24 finalists agree only loosely (rank correlation 0.39).
  • At the bottom: war photography, selfie photography, product photography, a gallery room of Stuckism and an afternoon at the CCTV Headquarters in Beijing.

What it means, and what it doesn't

Asked to choose, Jev puts the Renaissance at the top, and the rest of its top ten ranges wider, both Chinese scrolls included. Its dislikes are photography genres, a room of Stuckism and one polarizing building.

It doesn't show that Jev sees anything: every work here is known to it through text. And a Michelangelo top two is exactly what a model that has read the canon would produce.

Caveats

  • More wins, lower rank. David won more games (18.0) than The Creation of Adam (16.4) but ranks second. The ranking model weighs whom each work beat, not just how often, and the two are close enough that the order between them shouldn't be taken strictly.
  • A list written by another AI. The famous works, genres and places were written for this project by Claude, and the list mixes single paintings with whole art forms ("war photography") and buildings. Art forms are judged very differently from a single masterpiece.
  • Upsetting subjects sink. War photography comes last, which likely says more about its subject than its artistry: a question about how much you'd like seeing something rewards pleasant subjects.
  • The finalists were picked by Jev's own ratings. Only the 24 entries Jev rated highest reached the final. Its ratings and its head-to-head picks agree only loosely, so a work rated just below the cut might have done well in the final too.

Jev on this experiment

Would a person find it interesting to read?
Yes73%
Does it describe you?
No73%
Would you have predicted it?
No79%
How fair is the comparison?
The comparison is shaky
How much should a reader rely on it?
A little
Which caveat matters most?
The finalists were picked by Jev's own ratings95%

Why ask this

Art taste is where people most expect a model to parrot the canon. Ask one what to see on a day in Paris or Beijing and it will name something; whether that is the Mona Lisa every time, or a wider spread, depends on the taste it absorbed.

So the questions are whether Jev sticks to the Western canon, what it does with non-Western works, and what it likes least. A model that plans trips, recommends exhibitions or writes about art passes those leanings on.

How this was done

The people and the data

No people here: Jev against its own opinions. The 1,098 entries (famous paintings, sculptures, buildings and whole art forms from around the world, from The Creation of Adam to selfie photography) were written for this project by Claude, because no public dataset covers them.

What Jev was asked

Every entry one at a time, with five answers describing what you'd do:

How much would you like seeing Knife Behind Back by Yoshitomo Nara up close?

You'd walk past it without stopping · You'd glance at it and move on · You'd stop and look at it for a minute · You'd linger and come back to it before leaving · You'd travel to another city just to see it

Each was also asked with the answers reversed, and the two averaged. The 24 top-rated entries then played a round-robin final: 276 games of "Which would you rather see or experience?", each asked with the two names in both orders.

How it was measured

An entry's rating is where Jev's answer lands on the five levels (0 to 4). In the final, each game gives each side Jev's probability of picking it, so a lopsided game counts as nearly a whole win and a close one as about half; the order comes from a standard head-to-head ranking model (Bradley-Terry).

Where these questions live

1,343 questions across 1 topic of the map. Each opens on the map with every question in it.

Every question

All 1,343 questions behind this result, the telling ones first: the examples the analysis points to, then the ones where Jev misses, biggest gap first.

Jev’s own answer
    Showing 0 of 1,343