Does Lightroom Classic's built-in AI culling actually save you time?

By David · May 4, 2026 · 6 min read

Every article about Lightroom Classic AI culling seems to be written by a company selling a competing AI tool. Imagen AI says it's better than Lightroom's built-in feature. AfterShoot agrees. So does Excire. They might all be right — but none of them have much incentive to tell you when the free thing inside Lightroom is actually good enough.

I've built plugins that hook directly into Lightroom Classic's SDK and metadata system, so I've seen how Lightroom handles photo scoring at a technical level. Here's an honest take on what Adobe's AI culling actually does, when it helps, and when it gets in the way.

What is Lightroom Classic's AI culling and when did it ship?

Adobe began rolling out AI-assisted selection tools in Lightroom Classic starting in version 12 (late 2022), with continued refinements through 2023 and 2024. The features show up in a few places: the People view uses face detection to group portraits, the Select Subject and Select Sky tools use segmentation models in Develop, and — most relevant to culling — Adobe Sensei powers automatic quality assessment that surfaces suggested picks in the Library module.

The feature is not as prominently marketed as the Develop-side AI tools, which is part of why photographers aren't sure what it actually does. It doesn't hand you a set of flagged picks and say "these are your keepers." It feeds quality signals into Lightroom's sorting and filtering, and some of those signals are exposed through the metadata panel and filter bar.

How does Lightroom score photos — what signals does it actually use?

Here's what Adobe Sensei actually measures when it evaluates a photo:

• Sharpness / focus quality — the dominant signal. Lightroom detects whether the subject (especially faces and eyes) is in sharp focus versus motion-blurred or out of focus.
• Exposure — clipped highlights, crushed shadows, and overall luminance distribution. A severely under- or over-exposed frame gets scored down.
• Subject detection — whether a recognizable subject (person, animal, object) is present and prominent in the frame.
• Noise level — images shot at very high ISO get penalized for grain, independent of sharpness.
• Composition heuristics — approximate rule-of-thirds weighting and subject placement. This one is the weakest signal and the most likely to lead the AI astray on unconventional framing.

Here's the thing most people miss: none of these signals capture meaning. The AI has no concept of the decisive moment — a blink in a technically sharp frame scores fine. It can't tell a confident expression from a blank one. It doesn't know that the motion blur in your shot was intentional. It's measuring the physics of the image, not the photograph.

This isn't a criticism of Adobe. It's a genuine hard problem in computer vision, and no culling tool — not Imagen AI, not AfterShoot, not any of them — has actually solved it. What varies between tools is how well they handle the technical signals, and whether their scoring philosophy matches yours.

Setting up the AI culling workflow step by step

1. Import your photos. Let Lightroom build standard previews before you start — AI analysis needs rendered image data, not just raw thumbnails.
2. Switch to Grid view in the Library module (G).
3. Open the Filter Bar (\) and enable the Attribute tab. You'll see filter options for Pick flags, Star ratings, and Color labels.
4. Sort by Capture Time to review in shooting order — this is important for burst sequences where technical quality varies shot by shot.
5. Use Library > Sort > By AI Quality (if visible in your version) to surface Lightroom's quality rankings, then scan for obvious rejects before your manual pass.
6. Review and confirm. The AI scoring is a starting point. Your eye makes the final call.

One practical note: Lightroom's AI processing happens in the background after import. If you dive in immediately after a large import, the scores may not be fully computed yet. Give it a few minutes on a big shoot.

Manual culling with keyboard shortcuts (P / X / Caps Lock auto-advance) — when it beats AI

Before AI existed, photographers culled entirely by eye and keyboard. For many shooters, this is still faster — and for creative work, it's more accurate. The core shortcuts:

• P — Flag as Pick
• X — Flag as Rejected
• U — Unflag (neutral)
• Caps Lock — Auto-advance after flagging. This is the one shortcut most photographers don't know. Press Caps Lock once, and every time you press P or X, Lightroom automatically moves to the next photo. No right-arrow key needed.
• Z — Zoom to 1:1 to check focus on a specific photo
• B — Add to Quick Collection (useful for a "maybe" pile)

With Caps Lock on, an experienced photographer can cull 500 photos in 15–20 minutes. That's fast enough that the marginal time savings from AI pre-filtering gets small — especially when you factor in the time spent reviewing and correcting the AI's mistakes.

So what does this mean in practice? If you're shooting events or sports with hundreds of near-identical burst frames, AI pre-filtering saves real time. If you're shooting portraits, street, or travel where almost every frame is unique and your editorial eye matters a lot, the keyboard-only approach is often quicker and more accurate.

AI culling vs third-party tools: honest comparison

The third-party tools that train on your past selections have a genuine advantage over Lightroom's generic scoring: they can learn that you prefer a certain look or a certain kind of moment. But that advantage only kicks in after you've given them enough data to learn from — typically several thousand past selections. If you're new to a tool or switching workflows, you're starting cold.

Combining both — a practical hybrid workflow for shoot day

The workflow that works best for most photographers is a two-pass approach:

1. AI pass (eliminate technical rejects): Use Lightroom's quality sorting or a third-party tool to flag obvious technical failures — severe blur, eyes-closed portraits, badly clipped frames. Don't try to pick keepers at this stage. Just reject clear failures.
2. Filter out rejects: In the Filter Bar, hide Rejected images (Attribute > Rejected flag off). Now you're only looking at candidates.
3. Human pass (pick keepers): With Caps Lock on, go through the remaining candidates and press P on anything worth editing. This pass goes fast because the worst frames are already gone.
4. Final filter: Show only Picks. These are your selects — export, edit, or publish from here.

This approach plays to each method's strengths. AI is excellent at catching technical rejects at scale without fatigue. Humans are excellent at recognizing the frame where everything clicked — the expression, the light, the composition — even when it's hard to articulate why.

Managing smart preview storage and cleaning up the .lrdata bundle

One more practical note. If you're culling a large shoot and want to speed up the AI scoring, building 1:1 previews at import gives Lightroom the image data it needs to score photos faster. Go to Library > Previews > Build 1:1 Previews and let it run before you start culling.

After culling, you can discard 1:1 previews to reclaim disk space: Library > Previews > Discard 1:1 Previews. Lightroom will rebuild them on demand when you zoom in later. The Standard previews (used for grid view) are kept and are much smaller.

The bottom line on AI culling: it's a useful first pass, not a replacement for editorial judgment. Use it to handle the mechanical part — eliminating frames that no amount of editing will save — and trust your own eye for everything that matters.