ways of [machine] seeing

This is a website for a growing collection of activities for Art & Design teachers to share ideas which explore current developments in image-based Generative AI. Our interest is how both humans and machines are taught to see, in turn how visual culture is being transformed by developments in computer vision.

We take inspiration from John Berger’s Ways of Seeing to provoke some critical questions and creative pedagogy.

[pre-activity notes]

[before starting the activity]

Compliance: Check your school’s rules with regards to the use of AI and devices in class. Ensure your intended activity is adapted to conform to your school rules.

Test: Run the activity yourself. Ensure there are no technical issues. Seek advice from an IT technician to overcome issues such as firewalls and wifi connection.

Accessibility: Do not assume that everyone will have a device or be able to use a device easily. Develop an understanding of digital literacy amongst your students.

Review glossary: Familiarise yourself with key terms by reading through the glossary. You might want to do this together in class.

Check privacy settings: Make sure any AI tools used comply with your school’s data and privacy policies.

Plan for discussion time: Leave room to reflect and talk about what the AI got right, wrong, or what was surprising. The activities include questions to help facilitate this discussion.

Four colorful objects are paired with incorrect labels: a green horse head labeled "the door," a purple clock labeled "the wind," a red vase labeled "the bird," and a blue briefcase labeled "the valise."

[activities]

A yellow magnifying glass is partially enclosed by a grey bounding box labeled "pizza".

The Sleep of Reason

Exploring differences between how machines and humans ‘see’ an artwork.

An orange pencil and lined paper are partially enclosed by a grey bounding box labeled "fork".

Inside the Black Box

Drawing diagrams to imaginatively explore what is happening inside the black box of AI.

A red robot head is partially enclosed by a grey bounding box labeled "hand".

Make Like a Machine

Sculpting objects with clay to explore textual description, interpretation and how bias arises.

A pink house with surrounding bushes and a tree is partially enclosed by a grey bounding box labeled "mountain".

(Mis-)Representing Place

Exploring personal and algorithmic representations of place through visual research.

Two purple theatrical masks, one smiling and one frowning, are partially enclosed by a grey bounding box labeled "leaf".

Unmasking Facial Recognition

Creating masks to obfuscate the ways that machines recognise faces to understand how they ‘see’.

Two blue puzzle pieces are shown, partially enclosed by a grey bounding box labeled "ladybird".

Assembling Art History

Creating object-collages to explore collective understanding of art history and how this compares to what AI knows.

A blue artist’s palette with paint blobs is partially enclosed by a grey bounding box labeled "nose".

Order in the Art Room

Bringing order to the classroom to make it machine-readable, and exploring the art room and its objects as a dataset.

A blue chair is partially enclosed by a grey bounding box labeled "gate".

One & (More Than) Three Chairs

Exploring different representations of a classroom chair, from its close observation in text and image to new forms using AI.

A green lightning bolt is partially enclosed by a grey bounding box labeled "rocket".

Environment Matters

Thinking about the environmental impact of different technologies in the classroom, and introducing the idea of carbon literacy.

[post-activity notes]

[at the end of the activity]

Reflection: Continue to discuss what came out of the activity, student engagement and how the activity might be improved. Make a note of what went well and what did not. Use this to inform future activities.

Address misconceptions: Clarify anything the students misunderstood, e.g., AI doesn’t “think”, have imagination, or have emotions. Encourage critical reflection.

Document the activity: Capture process and outcomes, reflections or examples to use in future lessons and to share with colleagues.

Feedback: Collect feedback from students either formally or through observation. We are keen to hear your thoughts on the activities and any ideas or documentation you may have. Please use the ‘contact us’ form below.

A blue rocket is partially enclosed by a grey bounding box labeled "ball".

This project is [supported] by the Engineering and Physical Sciences Research Council [grant number EP/Y009800/1], through funding from Responsible Ai UK [RAI-SK-BID-00071].