Art Collector Pulls Focus with Sam Leach

Summary

In this Art Collector Magazine interview, artist Sam Leach discusses his exhibition 'Fully Automatic' at Sullivan and Strom in Sydney. Leach uses machine learning to generate imagery that intersects with art history and contemporary aesthetics. His work explores utopian visions of a post-scarcity society, inspired by 60s architecture and the book 'Fully Automated Luxury Communism'. The process involves curating a dataset of historical and contemporary images, which the algorithm then synthesizes into new compositions. Leach finds the algorithm's output both a framework for his paintings and a source of creative inspiration, leading to a deeper engagement with his art.

Takeaways

• 🎨 The artist Sam Leach discusses his work featured in the 'Fully Automatic' exhibition at Sullivan and Stroom in Sydney.
• 🤖 'Fully Automatic' is an ironic title, reflecting the use of machine learning to generate imagery, which is part of an art historical and contemporary dialogue.
• 🖌️ Sam's practice often draws from art history, contemporary science, and architecture to find aesthetic parallels over time.
• 🔍 Machine learning was employed to automate the process of finding aesthetic parallels in image archives, which Sam then translates into paintings.
• 🛋️ The artwork 'Beshero and Super Studio' combines elements of Super Studio sofas with people absorbed into them, referencing Boucher paintings.
• 📚 The title 'Fully Automatic' is inspired by Aaron Bastani's book 'Fully Automated Luxury Communism', proposing a utopian vision of a post-work society.
• 🎭 Sam is interested in utopian visions of the 60s and 70s, particularly the idea of a luxurious sofa in a post-scarcity society.
• 🤝 There is a conversation between the algorithm's aesthetics and Sam's preferences, with Sam tweaking the algorithm to steer it in certain directions.
• 🖼️ The algorithm produces low-resolution images, requiring interpretation and puzzle-solving to translate into a detailed painting.
• 🎭 The process has not saved time but has allowed Sam to spend more time painting and less on the Photoshop part of his process.
• 🌄 Sam is exploring the possibility of using machine learning to reconcile early colonial Australian landscape paintings with their original scenes, potentially removing colonial filters.

Q & A

What is the title of Sam Leach's exhibition?

The title of Sam Leach's exhibition is 'Fully Automatic'.

What is the irony in the title 'Fully Automatic'?

The irony in the title 'Fully Automatic' is that despite the use of machine learning to generate imagery, the process hasn't saved Sam any time in his artistic process.

What is the significance of the painting 'Beshears and Super Studio'?

The painting 'Beshears and Super Studio' is significant as it is part of Sam Leach's body of work that explores aesthetics across time using machine learning.

How does Sam Leach use machine learning in his art?

Sam Leach uses machine learning to automate the process of finding aesthetic parallels in art history and contemporary imagery, generating a series of images that he then selects and translates into paintings.

What is the inspiration behind Sam Leach's interest in utopian architecture from the 1960s and 70s?

Sam Leach's interest in utopian architecture from the 1960s and 70s is inspired by the book 'Fully Automated Luxury Communism' by Aaron Bastani, which proposes a post-work, post-scarcity society powered by AI and automation.

How does Sam Leach's use of machine learning relate to his previous artistic process?

Sam Leach's use of machine learning allows him to spend less time on the Photoshop part of his process and more time on actual painting, though it also introduces a new challenge of managing databases.

What is the role of the algorithm in Sam Leach's artwork?

The algorithm in Sam Leach's artwork serves as a tool to process and find aesthetic parallels in a dataset of images, producing a series of images that Sam can then select and use as a basis for his paintings.

How does Sam Leach's artwork reflect the aesthetics of the algorithm?

Sam Leach's artwork reflects the aesthetics of the algorithm by incorporating elements that the algorithm identifies as aesthetically similar, often resulting in a blend of historical and contemporary visual elements.

What is the original intention behind Sam Leach's use of machine learning in his art?

The original intention behind Sam Leach's use of machine learning was to predict what his next painting should be, based on the trajectory of his past work, thus reducing the effort in conceptualizing and composing new pieces.

What future projects is Sam Leach considering in relation to his use of machine learning?

Sam Leach is considering projects that explore the latent space between historical and visual culture, particularly in landscape painting, with the aim of reconciling early colonial Australian landscape paintings and reconstructing the original scenes.

Outlines

00:00

🎨 Artistic Integration with Machine Learning

In this segment, Louise Martin Shoe from Art Collector Magazine interviews artist Sam Leach about his exhibition 'Fully Automatic' at Sullivan and Strom in Sydney. The title is derived from Aaron Bastani's book advocating for a post-work, post-scarcity society powered by AI and advanced technologies. Sam discusses his use of machine learning to generate imagery that merges art historical references with contemporary themes, specifically utopian architecture from the 1960s and 70s. The process involves creating a dataset of images, which the machine learning algorithm uses to find aesthetic parallels, producing a series of images from which Sam selects to create his paintings. This approach has allowed Sam to focus more on painting rather than image selection and composition, although it has introduced new challenges in managing databases.

05:01

🤖 The Algorithmic Art Creation Process

This paragraph delves deeper into Sam Leach's creative process with machine learning. He explains how the algorithm generates low-resolution images that he interprets and translates into his paintings, comparing it to solving a puzzle. Sam discusses the enjoyment he finds in this process and how the algorithm provides a basic compositional framework for his paintings. He also talks about curating his own dataset for the machine learning process, focusing on images that have elements he wants to incorporate into his work. The goal was to predict and automate the creation of his paintings, although it did not entirely achieve this, it opened up new possibilities for exploring the latent space between different areas of history and visual culture.

10:01

🌄 Future Projects: Reimagining Australian Landscapes

In the final paragraph, Sam Leach hints at future projects where he intends to explore the latent space between historical and visual culture further, specifically focusing on early colonial Australian landscape paintings. He is interested in using the machine learning process to reconcile these paintings with the original scenes they were inspired by, potentially reconstructing what those scenes might have looked like without the colonial perspective. This approach aims to offer a new way of looking at Australia's early history, removing the colonial filter and offering a fresh perspective on these landscapes.

Keywords

💡Fully Automatic

The term 'Fully Automatic' is the title of Sam Leach's exhibition at Sullivan and Strom in Sydney. It is an ironic use of the title, as it suggests a process that is entirely automated, yet the artwork involves a combination of machine learning and manual painting. The exhibition explores the intersection of technology and art, particularly in how machine learning can generate imagery that fits within both contemporary and historical art contexts.

💡Machine Learning

Machine learning is a subset of artificial intelligence that involves algorithms learning from data. In the context of this video, Sam Leach uses machine learning to generate imagery by feeding it a dataset of images from art history and other topics of interest. This process allows him to find aesthetic parallels and create new compositions, which he then translates into paintings. It's a key component of his artistic practice, demonstrating how technology can influence and enhance traditional art forms.

💡Aesthetic Parallels

Aesthetic parallels refer to the visual similarities or connections that can be drawn between different artworks or elements within art. Sam Leach discusses how he uses machine learning to identify these parallels across time, drawing from art history and contemporary sources. This approach allows him to create a diachronic view of aesthetics, showing how certain visual elements or themes recur or evolve in different periods.

💡Utopian Architecture

Utopian architecture is a concept that envisions ideal living spaces and environments, often characterized by harmony, efficiency, and beauty. In the video, Sam Leach mentions his interest in utopian architecture from the 1960s and 70s, which he incorporates into his artwork. This interest is tied to the broader theme of a post-work, post-scarcity society, as proposed in Aaron Bastani's book 'Fully Automated Luxury Communism,' which inspired the exhibition's title.

💡Boucher and Super Studio

Boucher and Super Studio are references to specific art historical contexts that Sam Leach draws upon in his work. Boucher refers to François Boucher, a French painter known for his Rococo style, while Super Studio is a reference to the Italian architectural group active in the 1960s and 70s. These references are used in the machine learning process to generate imagery that combines elements of historical painting with contemporary architectural ideas.

💡Post-Scarcity Society

A post-scarcity society is a theoretical concept where resources are abundant, and there is no need for competition or struggle over basic needs. Sam Leach discusses this idea in relation to Aaron Bastani's book, which proposes a future where artificial intelligence and automation provide for everyone's needs. This utopian vision is reflected in the themes of Leach's artwork, particularly in the depiction of luxurious and leisurely lifestyles.

💡Data Set

A data set in this context refers to the collection of images and information that Sam Leach uses to train his machine learning algorithm. The artist carefully curates this data set, selecting images from art history and other sources that he believes will help the algorithm identify aesthetic parallels. This curated approach allows him to guide the creative process and ensure that the generated imagery aligns with his artistic vision.

💡Conceptual Art

Conceptual art is an art movement that emphasizes the importance of the idea or concept behind the artwork, rather than the physical object itself. Sam Leach mentions his interest in the conceptual aspects of his work, suggesting that the themes and ideas he explores are as important as the visual elements. This approach is evident in how he uses machine learning to generate imagery that reflects on historical and contemporary art practices.

💡Latent Space

Latent space is a term used in machine learning to describe the underlying, often abstract, representation of data that the algorithm uses to make predictions or generate new content. In Sam Leach's work, exploring the latent space between different areas of history and visual culture allows him to create new compositions that blend elements from various sources. This exploration is a key part of his creative process, enabling him to discover new connections and possibilities.

💡Early Colonial Australian Landscape Paintings

Early colonial Australian landscape paintings refer to artworks created during the period of European colonization of Australia, often depicting the natural environment and the impact of human settlement. Sam Leach expresses interest in using his machine learning process to reinterpret these paintings, potentially reconstructing the original scenes they were based on. This approach could offer a new perspective on Australia's early history, removing some of the colonial bias that may have influenced these early depictions.

Highlights

Sam Leach discusses his exhibition 'Fully Automatic' at Sullivan and Stom in Sydney.

The title 'Fully Automatic' is ironic, reflecting on the use of machine learning in art creation.

Leach's practice combines art historical and contemporary imagery to find aesthetic parallels.

Machine learning automates the process of finding aesthetic parallels in art history and utopian architecture.

The algorithm generates hundreds of images for Leach to select from for his paintings.

The computer algorithm sometimes directly quotes historical paintings, other times it blurs or alludes to shapes and forms.

Leach's work combines elements of Super Studio sofas with figures demonstrating them, absorbed into the sculptures.

The exhibition title is inspired by Aaron Bastani's book 'Fully Automated Luxury Communism'.

Bastani's utopian vision of a post-work society supplied by AI and 3D printing inspires Leach.

Leach is interested in the utopian visions of the 1960s and 70s, especially related to luxurious living.

The algorithm and Leach have a conversation through the process of generating and selecting images.

The process of translating algorithm-generated images into paintings is likened to solving a puzzle.

Leach curates his own collection of images to train the algorithm, ensuring specific aesthetic outcomes.

The original intention was for the algorithm to predict Leach's next painting based on past works.

The algorithm opens up possibilities for exploring the latent space between areas of history and visual culture.

Leach plans to use the process to reconcile early colonial Australian landscape paintings with their original scenes.

The project aims to provide a new perspective on early Australian history, removing the colonial filter.

Leach sees potential in using machine learning to reinvent the past with a futuristic approach.

Transcripts

hello so it's louise martin shoe here

i'm with artist sam

leach for art collector magazine and

we're discussing one of the paintings

that features in his exhibition which

just opened

at sullivan and strom in sydney and the

exhibition's called

fully automatic which is a really ironic

use of the title from what i understand

so i'm speaking to you from outside

brisbane and sam's lockdown

in melbourne hi sam great to see you

hi louise good to see you too so sam in

this

pull focus art collector series we

discuss one particular work

and we've chosen beshear and super

studio which is from 2020

and that's part of this really

interesting body of work in

um exhibited under the title fully

automatic

so um just to go straight to the

question sam

fully automatics an experiment as i

understand it using machine learning

to generate imagery that as i understand

it kind of fits within your own move

but also within an art historical canon

so i'm just going to share this screen

so we can look at this

work and maybe sam you could talk to us

about how we see that manifest in

boucher and super studio

yeah so um basically my

my practice uh has very often drawn on

art historical uh as well as you know

contemporary

scientific and architectural um imagery

to um i guess try and draw out aesthetic

parallels

over time sort of a diacronic approach

to aesthetics and

um you know what i find interesting is

sort of drawing out from the aesthetics

aspects of of culture that are reflected

visually in those

in those elements so the machine

learning was a way

of me um in a sense trying to automate

that

process so normally i spend a lot of

time

going through archives of of images from

various sources and looking for these

for these parallels and trying to

put them together into an image which i

then translate

to a painting but with this i can

basically put together a whole data set

of images

from from art history plus whatever

topic i'm interested in in this case

uh utopian architecture ideas of the

1960s and 70s

and let the let the algorithm let the

the machine learning itself

do that do that processing of finding

the aesthetic parallels and and pulling

it together into an image

and it'll basically produce um a series

of hundreds of images and i can just

sort of scroll through and select from

the ones that i think look interesting

to to generate into a painting which is

which is how we sort of arrive at this

image

so the computer's not really the

algorithm rather is not really

in some instances it's sort of directly

quoting from from the historical

paintings and in some instances it's

sort of more of a

a blurring or a smudging or just an

allusion to a particular

shape or compositional form that that's

appearing

so in this case um it's actually

uh i guess combined elements of

these super studio sofas with the uh

people who are who are sort of

demonstrating them so

there's people that have sort of been

absorbed into these into these

sculptures

uh these sculptural sofas uh as well as

the the form of the sofa itself

um and it's sort of found there's an

allusion to the

boucher paintings which you would find

you know these figures kind of reclining

on clouds floating into the floating

into the sky

um and so it sort of pulls them into you

know to me this quite satisfying

melange of these of these uh factors

so what kind of stimulated the

the ai and you to draw together bushi

and

super studio is it because of your

interest in the 60s

architecture or um yeah well

in particular um with this show

that the title fully automatic comes

from a book by aaron bastani fully

automatic luxury communism

where he's basically proposing a

manifesto for a post-work post-scarcity

society that's being supplied by

artificial intelligence and 3d printing

and

you know automatic robotic processes and

you know i i love this utopian vision

you know his his kind of catchphrase

infinity pools for everyone

and i think yes you know that sounds

that sounds great well you know it's uh

it's communism with champagne i think

this sounds it sounds awesome

so i really so that's why you know i'm

very interested in the utopian visions

of

um of the 60s and 70s you know that kind

of ideal society especially when it

comes to things like a really luxurious

uh sofa that you know you can imagine

just sprawling out in in your in your

post-scarcity life

and the model um and actually um the

model he says for a society that that

is like this in case people are worried

they're going to get bored is really you

know the aristocracy of europe

in the uh you know in the in the

enlightenment period so

those paintings by you know fragonard

and boucher and others

you know that rokoko period really um is

an illustration of this kind of utopian

utopian life so it's a natural it's a

natural combination

to me the other aspect is that when i'm

programming

the uh the algorithm when i'm supplying

it it's it's

if you like drawn to certain uh

graphical images

from the data set and so i kind of i

kind of respond to that and tweak it by

adding images or weighting certain

things slightly differently to

to just steer it in in certain

directions to produce more interesting

results for me

so there's a little bit of a little bit

of a conversation between what the

aesthetics of the algorithm are

if you like and and my own preferences

so that also

is a it comes into play so there's a

conceptual aspect plus

uh you know just what the maths wants to

do and

fully automatic is an ironic title

because this process hasn't really

saved you any time as i understand it

um no yeah that's right yeah it's but i

mean i guess it's uh

i mean it's worked in a way because what

what i wanted to do was

um spend less time on you know really

the

the photoshop part of my process where i

would

buy these images and spend more time in

the actual painting and it's absolutely

done that

um but it does take me a lot longer to

do the paintings

and also i've discovered a whole new

body of work in managing databases so

that's

an exciting development and can you just

step us through what the journey is

between what the computer

kind of generated for you and what we

see

in the image that you've painted

so what what i'm looking at from from

the algorithm it's

it produces uh i guess a a fuzzy

often fairly pixelated image and it's

quite it's quite low resolution so

it's only about um 128 pixels by 128 so

you know really almost thumbnail size so

there's a lot of

um interpreting if you like of what um

you know what's going in between those

pixels when it comes to translating it

to

to a painting but that's actually you

know a really enjoyable part of it

that's like

like figuring out a puzzle where you've

just got these you know these two points

on a canvas that you need to connect

somehow and what

like what is the thing that connects

connects those two so that's been

that's been like a really enjoyable part

of of that process

uh but it's really nice also to have

what this this algorithm is producing

it's like a framework

just the just the basic uh composition

that i can fall back to so when i'm

thinking

well what happens up in this corner i

can just look at this thing on the

algorithm and say okay well

uh you know there's some kind of purple

action happening up there so i'll start

with the purple action and then

uh you know resolve it back so there's

these little hints uh and guides that

i'm getting from the algorithm and then

the rest is

is happening on the canvas while i'm

painting it such an interesting process

because i guess if you're feeding in the

historical canon you've got a lot of

already successful

pictures so do you think this

predisposes it

to produce a successful composition

yeah absolutely it does it does do that

and um you know of course

i'm like i'm in control of what's going

into that data set so a lot of

a lot of what's been happening in this

space with machine learning

is using large publicly available uh

data sets you know taken off off wiki

art or you know celebrity

portraits or things like that but i'm

i'm really you know curating curating my

own collection

that uh that i'm training it on so i'm

quite specific about the images so

not only yes there are paintings that i

consider to be

successful um but they're paintings that

have some element that i want to

um draw from in in producing my own work

um whether that's you know the

conceptual content because i you know i

do always like to

um you know have some fun with the with

the conceptual stuff in the in the works

but also just the just the technical

aspect of the paintings and how that how

they put together the things that i like

to see in there

and did this process meet your original

intention for the painting and for the

project i mean

is this an experiment you'll continue to

work with

oh yeah definitely i mean it's so

there's so much to do with it

um actually the i'll answer the that's

so that's answering the second

part first the first part the original

intention

was to um have a way to uh

predict what my next painting should

should be so

i had this idea that i would basically

turn up in the studio in the morning

um hit enter on the computer and say

okay well today

make this painting you know based on all

the paintings that i've done in the past

it would just look at a trajectory and

make a prediction

and then i wouldn't have to do any of

the hard work of of dealing with

concepts or composition or anything i

just

get a print and i'll color it in and i

thought that that sounds great

so um no it didn't it didn't quite

it didn't quite do that um but that is

still

that is still a project that i'm that

i'm trying to develop

um but it did open you know so many so

many fascinating possibilities

because you know what it's really doing

is exploring this latent space between

various areas of history and various um

aspects of visual culture and it does it

in a way

that i certainly wouldn't do

without the prompt of this of this

algorithm so

it's really giving me a new uh angle

to to look at these these uh you know

these archives that i've been looking at

for some time which

which is something i really enjoy it's

an intriguing project we could talk

about it for um

hours but in the time we've got um is

there anything else that you think

collectors will be interested to hear

about with your upcoming projects sam

well one of the things that i'm i'm

interested in in looking at

in relation to that latent space is

particularly with landscape painting is

looking at

early colonial australian landscape

paintings and

with this process it should be possible

to reconcile

landscape paintings with the imagery

that they're

taken from and then feed back into it

paintings where we don't necessarily

have an image of the original scene

but have it reconstruct what that

actually would have looked like so

um i feel like it's it's a possibility

to give us a new um

a new way of looking at this this early

history of australia and maybe try and

take off some of that colonial filter

that's that's been applied to it

so you're kind of reinventing the past

with the future

yeah yeah yeah yeah yeah exactly yeah

um look thank you so much sam's been

really fantastic to speak to you

um and i'm sure all of our viewers would

be really interested to see

fully automatic which is at sullivan and

strength in sydney until the 12th of

september

thank you thank you very much