What happened
Which?, the UK consumer organization, published a comparison of the AI features of three flagship phones on 1 August 2025. Its testers ran the Apple iPhone 16 Pro Max, the Google Pixel 9 Pro XL and the Samsung Galaxy S25 Ultra side by side in its labs. One test searched each phone's photo gallery with natural language, such as the phrase 'relaxing on a boat'. Which? found that the iPhone was excellent at this. The phone located photos by objects, descriptions, text within a picture and landmarks, even when a photo had no location data. The Pixel was often less accurate, and Samsung's gallery could not find text within images or landmarks. The iPhone was also the only phone that could find a moment within a video, such as when a fish appeared in an aquarium clip. Apple describes natural language search in Photos as part of Apple Intelligence. Which? gives no counts of searches or error rates.
Sources
-
Assessment report, by authors with no tie to the system.
Which? (Adam Speight). Apple vs Google vs Samsung: which phone has the best AI? 1 August 2025. Which? tested the phones' AI features side by side in its labs. Section on photo, video and image tools, natural language gallery search.
The iPhone 16 Pro Max was excellent at this, successfully finding photos based on objects, description text (such as the phrase 'relaxing on a boat'), text within a picture and even landmarks when the photo had no location data attached.
The iPhone's gallery search was found to be superior, making it much easier to find specific photos by searching for descriptive text or even landmarks. It was also the only phone that could find moments within a video.
-
Press release, by the makers.
Apple. Apple Intelligence is available today on iPhone, iPad, and Mac. Press release, 28 October 2024. Quotation from Craig Federighi, Apple's senior vice president of Software Engineering.
Apple Intelligence unlocks exciting new capabilities that make your iPhone, iPad, and Mac even more helpful and useful, from Writing Tools to help refine your writing, to summarized notifications that surface what's most important, to the ability to search for almost anything in your photos and videos by simply describing it
Details
AI system
- Workflow
-
Which? Searched each phone's photo gallery with natural language in its labs. The search ran in the gallery app because none of the phones let testers start it from the main voice assistant.
- Relation to AI
-
Which? Tested the search directly, so the results that it judged were the feature's own output. It compared three phones' gallery searches with one another and did not time searches or compare them with browsing by hand. The article does not describe the model. Apple's release presents photo search as part of Apple Intelligence, which it calls generative AI.
Follow-up
- Conditions for replication
-
The phone must support Apple Intelligence, which Which? Lists for all iPhone 16 models and the iPhone 15 Pro and Pro Max. Which? Reports that Apple says that some features may not be available in all regions and languages.
Entry details
- Sector
-
Consumer technology (photo search).
- Purpose
-
Apple built natural language search in Photos, part of Apple Intelligence, to find photos and moments in videos from a description.
31 more questions not yet checked
model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). iPhone natural language search finds photos best in Which? tests (AIBD-1053). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-1053.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)