Robotics & Automation News

Where Innovation Meets Imagination

Sotheby’s auctions painting by a robot for $1.124 million

The New York branch of Sotheby’s, one of the world’s most prestigious auction houses when it comes to fine art and artefacts, has just sold a painting by a robot for more than $1.124 million.

The painting, called “AI God – Portrait of Alan Turing”, was produced by the humanoid robot Ai-Da, which readers of this website will probably know from our previous coverage, and which has its own website.

The estimated sale price of the mixed-media work was listed as being between $120,000 to $180,000 by Sotheby’s, which clearly did not accurately predict how much revenue the work would generate.

The buyer is not yet known, but they are not alone in their interest in art produced by or using new technologies.

According to ArtNews.com, Sotheby’s also sold another 16 works of art by “digital artists”, by which they probably mean human artists who use digital technologies, although we are not sure about that.

But of the total sales revenue for digital art of just under $1.5 million the majority was obviously generated by the sale of the painting by Ai-Da, which was itself created by roboticist Aidan Meller.

Meller, who also owns an art gallery in London and deals in art, said in advance of the sale that the auction marks “an important moment for the visual arts… as AI gains more power”.

This is the first time a work of art produced by a humanoid robot has gone on sale at an auction house as revered as Sotheby’s.

However, there are many AI engines currently online that can generate images from one single keyword. We found this one – Deep AI – quite good at producing interesting images from one word. We didn’t try its music or video generators, but we guess they are equally impressive.

All of which raises one of the fundamental questions about the nature of art. In other words, what constitutes art?

It was often said that robots might be good at repetitive tasks, but they are not creative. However, the voices that used to say that have become somewhat quieter. We used to say the same thing – we are basically talking about ourselves.

It’s been a few years since robots and artificial intelligence were shown to be capable of such things – painting works of art, editing video and making music – but now, they have become more widespread and more powerful.

One could still – correctly, in our view – equate a robot used by a human to a tool, much like a pencil or paintbrush. But they still have a level of autonomy – albeit within parameters set by the humans who create them – that will increasingly cause people to question what it is to be human.

What is it that makes us different from robots and AI?

We can feel confident that we are “different”, but we can’t be so confident that robots are not capable of being as skilful as we can be, maybe even as “learned”.

It was the late Alan Turing, the English mathematician and inventor of an early computer and the subject of Ai-Da’s painting, who asked the question, “Can machines think?”

We don’t believe they can.

But the test named after him, the Turing Test, is increasingly becoming untenable, especially as work produced by AI gradually permeates every part of the internet.

Pablo Picasso, one of the most celebrated (human) artists ever, is reported to have once said that computers can only provide answers – they can’t ask questions. Sounds good. We’ll buy that. Maybe therein lies our special quality that robots and AI do not yet have: we can wonder, and ask questions about, all sorts of things in life, the universe and everything, and investigate them in ways that robots and AI can only dream of. And that’s the other thing: robots and AI can’t dream – we can.

Having said that, AI is said by its proponents to “hallucinate”. But that term is used loosely and is meant to indicate that the AI is simply providing false or misleading information and presenting it as fact.

We don’t know where this is going, but whether we like it or not, what we might call “the human domain” in this context, appears to be getting smaller and smaller, or at least more difficult to differentiate from areas the robots and AI are increasingly taking over.