an interactive exercise

Visual Literacy in the Age of Machine Vision

Algorithms are constantly feeding us images designed to hold our attention for as long as possible, through outrage, fear, entertainment, and manipulation. That makes it more important than ever to understand what makes a human's reading of a photograph different from the way a machine registers it.

human

A human reads for meaning

When you look at a photograph, you're not decoding pixels — you're inferring a world. Who are these people, and how do they relate to each other? What just happened, or is about to? What's outside the frame? Does it make you feel something? You bring your own history to it, and no two viewers read it quite the same way. Ambiguity isn't a failure here; it's often the point.

machine

A machine reads for classification

A computer vision model never sees a "scene." It sees a grid of numbers, such as brightness and color values, converted into statistical patterns matched against categories it was trained on. It outputs whatever labels its training data taught it to associate with similar patterns. There's no ambiguity tolerance, no narrative, no nuance.

Why this gap matters

Most images made and processed today are never seen by a human eye at all.

About this exercise: what you will experience

This interactive exercise uses a photograph of your choosing to walk through both readings. First, a human's rich, ambiguous process. Then, a machine's flat, confident-sounding labels. And finally, a chance to compare the two. For best results, set aside about 10 minutes to go through it all. There's a lot to learn here!

The Process for Critically Understanding a Photograph

Dive into photography!

The sheer volume of images we interact with has made a real human connection with a photograph a rarity. Let's learn how to luxuriate in that process.

The habit to break

Glance → Evaluate

We all do it. Skim the surface, jump straight to a verdict, keep scrolling.

vs

The practice to build

Look! → D.I.V.E.

Concentrated and slow looking first. Then a real descent — four deliberate stages that lead to deep understanding.

Descend through the process ↓

SURFACE

Before D.I.V.E. even begins

Look

Not just a glance. Let your eyes linger on the details of the photograph with no goal except to actually see it. This is the plunge; everything below depends on it. ↓

−6m

DDescribe

What's actually there

What is the photograph of and how was it made? Subject, framing, vantage point, the moment of exposure, focus, light, color. What photographic choices did the maker employ? Describe the raw materials, meaning comes later. ↓

−13m

IInterpret

What it might mean

What's happening in the photograph, and what's the story or claim behind it? Form an interpretation by pointing back to the raw materials. An interpretation without evidence is just a guess. ↓

−20m

VVerify

Check your interpretation against the record

How do you know if you're reading it correctly? Ask the photographer, if you can. Read their statements. See what critics and scholars have written. And if none of that's available, the meaning you derive still has real value. ↓

−27m

EEvaluate

What it's worth

Does the photograph succeed on its own terms? Weigh its craft (studium) against what pierces you personally (punctum). ↓

Your turn

Choose carefully

Pick a photograph that means something to you, one you're willing to spend real time with. You'll go through the critical understanding steps, then see how a machine reads the same image.

Select a photograph or drag one here

step 1 of 2

Teach it to see you.

Move your mouse onto each dot and look right at it — hold still for about half a second and it fills in on its own. You can also just click it instead. Just like any machine model, this one only knows what you show it.

Your eyes are what it's actually learning from — keep them on the dot the whole time your cursor sits on it, not just at the moment you click.

Where you're sitting matters, too — try to stay in this same spot (same distance from the screen, same head angle) during Deep Looking afterward. Moving around will make the tracking less accurate.

deep looking

Ready to look?

When you click start, the photo will fill the screen and this page will quietly track where your attention goes. Take as long as you like — there's no timer or task.

For the most accurate results, stay in the same position — same distance from the screen, same head angle — that you were in during calibration. The red outline below shows where your face was then; try to line up with it before you start.

Not feeling well-positioned?

deep looking — results

You looked at the photograph for —.

You looked away for —.

Your attention heatmap:

This heatmap is a rough estimate from a consumer webcam, not the precise results that dedicated eye-tracking lab equipment can get. But it hints at where your attention gathered.

Verify / Research

Artist Intention vs. Viewer's Interpretation

A critical analysis doesn't have to end with your own reading. The next question to ask is: how do I know if I'm interpreting this photograph correctly? There are a few real ways to try to find out:

Ask the photographer directly, if you can.
Read anything they've written about it themselves — an artist statement, an interview.
Read what scholars, critics, or curators have written about the work.

Most of the time — especially with a photo you found and can't trace back to its maker — none of that is available, or you just don't feel like digging. That's completely fine. If your reading doesn't match what the photographer intended, the meaning you found still has real value. That's one of the most powerful things about photography!

Evaluate

Is this a "good" photograph?

Photography theorist Roland Barthes described two different ways we respond to a photograph: studium, the more studied, objective judgment that tends to be shared across viewers, critics, and curators and punctum, a deeply personal, subjective response that can't really be explained or argued for.

Here, you'll rate the photograph both ways.

Studium 5/10

How well does the photograph communicate what you interpret, with the aid of research, to be its intended idea and meaning?

WeaklyStrongly

Punctum 5/10

Setting the analysis aside — is this simply a photograph you personally connect with?

It's not my tasteI would hang this in my house

Machine Reading

What does the machine see?

This photograph is now being sent to a real computer vision model — the same category of system used in facial recognition, content moderation, and image search at scale. Your photo is sent securely for this one analysis and is never stored or saved.

Analyzing…

Comparison

How you saw it, how the machine saw it

Same photograph, two very different ways of "seeing" it. Yours came from looking, thinking, and feeling; the machine's came from statistical pattern-matching against its training data. Neither is more "true" than the other — but they're worth holding side by side.

The Subject

What is this literally a photograph of?

You

—

Machine (Cloud Vision labels)

The Interpretation

What's happening in this photograph, and what's it about?

You

—

Machine (3 AI models)

The Evaluation

How strongly does this photograph communicate its intended meaning?

You (Studium)

—