If Visual Intelligence is meant to tell you about unfamiliar things in front of you by providing context you wouldn’t necessarily get otherwise, then trying it out on some paintings seems like the perfect application. Paintings are a visual medium with all kinds of angles from which to look at them, but if you don’t have a degree in fine arts, it can be difficult to approach. After recently visiting London’s Tate Britain gallery, I tried to change that.
My mission was to find and see a specific exhibition in a gallery, hoping that Visual Intelligence would help make the trip an easy but informative day. Although this art is not to everyone’s taste, I and many others want to learn more about it. And maybe Apple Intelligence can help with that.
Map reading with Visual Intelligence
Upon entering Tate Britain, I did the obvious thing and looked for a map that would direct me to the artist I wanted to check out – British Romantic artist JMW Turner. Fortunately, this map was easy to read and decipher on my own, but I pointed my iPhone at it anyway. If this map was written in a different language or alphabet, I would definitely need help reading it, and that’s where I was hoping visual intelligence could help.
Fortunately for me, this map was oriented vertically, matching the portrait view expected by Visual Intelligence. After taking a picture, the iPhone quickly recognizes most (but not all) of the text on the sign. From here I had two practical options, either to get Apple Intelligence to explain the sign to me, or to pass it to ChatGPT via the Ask button.
(Image credit: Tom’s Guide)
Apple Intelligence offered the possibility to annotate the text, as well as extract the date of February 16, 2025 (the end date of the temporary exhibition). The summary could tell me the name of the gallery, some of the prominent artists in it, and a few places I could visit. All this is useful to know, but it will not help me get to Turner’s paintings.
Then I tried ChatGPT. Surprisingly, he couldn’t tell me exactly where I was, only that it was the British Art Museum and a map. Fortunately, I could ask more questions, so I asked where I could find the Turner exhibit. ChatGPT was able to point out that the paintings were in the lower right corner of the map, but he misread the room numbers and also failed to include all the rooms that were part of the exhibition. Not a huge mistake, but one that could have caused confusion if I hadn’t been able to read the map from the start.
Artist identification using visual intelligence
(Image credit: Tom’s Guide)
As you enter the exhibition, you come across one of Turner’s most famous portraits – a major target for Visual Intelligence. ChatGPT immediately identified it and provided the appropriate context.
I also tried the search function (actually a shortcut for Google Lens ) which also worked well to show me results for this particular image on the web. Tapping on one of these results opens a browser popup over the Visual Intelligence interface, which works fine, but I’d rather stay within Visual Intelligence or go straight to my full-featured browser app as opposed to this half-assed solution.
Visual Intelligence scaling limitations
When I first tried Apple Visual Intelligence, I criticized it for only being able to magnify a digital image. Even though the iPhone 16 Pro Max I was carrying had a great 5x telephoto camera, it didn’t support Visual Intelligence. And this fact especially annoyed Tate Britain.
The marks next to each painting are easy to read from a distance, but if you try to zoom in to get the whole panel in the Visual Intelligence viewfinder, you end up with a blurry mess.
(Image credit: Tom’s Guide)
I had to stand right up against the wall to get a clear image of the sign that Visual Intelligence could generalize, which it did well, albeit with too little detail for my taste. But if there was a barrier between me and the wall, or a worried guard wondering why I wanted to get so close to these 200-year-old masterpieces, that would be a problem.
Fact-checking with Visual Intelligence
One of Turner’s most famous paintings in this gallery is Norham Castle, Sunrise a stunning, almost abstract image of a castle in northeast England. I learned from a plaque nearby that it was based on an engraving Turner had originally made for the book, and not drawn from scratch. This seemed like an interesting fact, so I checked to see if Visual Intelligence could tell me about it.
(Image credit: Tom’s Guide)
The search function wasn’t much help, only providing image results with no ability to refine the query. Meanwhile, ChatGPT took two clues to correctly identify the painting and then told me about the castle it was based on when I asked.
Realizing that my question was probably too vague in this case, I then asked more specifically if the painting was based on a book, after which ChatGPT explained the connection between the print and the painting. I eventually got where I needed to go, but only because I knew what the end goal was in the beginning.
Recognizing Lesser Known Works Using Visual Intelligence
Although Google The search results are not as diverse as what you can get from ChatGPT, I found them to be 100% accurate unlike a chatbot. For example, looking at Turner’s lesser-known work, Plowing turnips near Slough the search function correctly found relevant results on the Internet. When I used the “Ask” button in ChatGPT, it misidentified the artist and then the painting, even when I said it was by Turner.
(Image credit: Tom’s Guide)
It demonstrates a major limitation of ChatGPT’s integration with Apple Intelligence—each chat is a completely separate interaction. If you do something like this through the dedicated ChatGPT app, it can take your previous messages into account when answering subsequent questions. But for regular iPhone 16 users who want to experience Visual Intelligence without subscribing to any extras, this means you’re starting from scratch with every new image you take, requiring you to explain things to the phone over and over again.
Sharing stories with Visual Intelligence
My favorite painting hanging in this gallery Regulus not just because it looks incredible, but because of the legend surrounding it – that Turner accidentally punctured the canvas while painting the sun because he was trying to make it so blindingly white.
(Image credit: Tom’s Guide)
This is exactly the story you hope Visual Intelligence can tell you about a painting, so I asked it to tell me about a painting. After a false start in which the setting of the painting was mistaken for Venice instead of Carthage, I eventually got the story I wanted by asking about the damage to the painting during its creation. Asking about general trivia or more generally about damage didn’t give me the story I wanted.
Explanation of connections with visual intelligence
The last test I gave Visual Intelligence was to explain why paintings by another artist, John Constable, were hanging in the same part of the museum. Google immediately identified the painting, but the links only showed results about the painting, which didn’t help me answer my question as to why it was hanging there.
(Image credit: Tom’s Guide)
Again, ChatGPT used an additional clue to identify the painting and its creator, but he was able to explain the temporal and stylistic connection between Turner and Constable, giving essentially the same explanation as the gallery gave on the plaque at the entrance to that room. .
A picture is worth a thousand tips
When I tested Visual Intelligence at Tate Britain, I found that the three main components that make up the feature – the way Visual Intelligence collects information, and the Google and ChatGPT-based cores that provide the actual content – are at quite different levels. competence. But together they can offer enough generally accurate information, not like having a museum curator in your pocket, but more like a know-it-all uncle who tends to misremember an art class he took in college until you correct him.
First, Visual Intelligence has proven itself to be easy to use and able to help without any other services, thanks to text recognition, summarization and the ability to integrate with other Apple applications. But it really should work with optical zoom cameras where possible, for times when you can’t get close to your subject. It would be helpful to offer more and broader options for how to explore images beyond the current two services it also connects to.
I find Google Search Results to be the most robust of the Visual Intelligence suite, but it’s also the least integrated. The fact that the results are often just images can be a hindrance, and it’s surprising given that a regular Google search in the web browser will happily provide an AI summary or a highlighted snippet of text to tell you what you’re looking for.
Finally, we have ChatGPT, which would benefit from a large context window to explain more things in a few snapshots rather than needing prompts over and over again. It might not happen to balance the ChatGPT server costs and Apple offering its services for free to iPhone 16 users, but it’s still on my wish list. Being more specific, or perhaps being able to add a written hint along with an image when you first ask about it, can also help narrow down the results to what you need more quickly.
As a learning tool, Visual Intelligence has shown great potential, and if Apple can tap into that potential with future iOS updates and hardware generations, iPhone owners could have one of the best learning tools around. But now, and perhaps for a long time to come, it’s still faster to look for a sign or ask the nearest expert to get an accurate description of a painting than to nudge ChatGPT again and again to determine the artist’s name.
More from Tom’s Guide