Botto, the “artistic” AI that earns millions

Who is Botto, this artificial intelligence which creates “artistic” works… which are found in the galleries of Sotheby’s and generate millions of dollars.

What is Art? This is a question that artists and philosophers have been trying to answer for centuries. And with the appearance of artificial intelligence, the debate seems far from over.

But in the meantime, with two or three prompt lines, it is possible to obtain a drawing from a generative AI bot. And one bot in particular specializes in “artistic” creation: Botto. And his creations bring in millions of dollars. Five, since its inception, to be precise. A total of 150 productions were sold at auction.

“Autonomous”

Botto is a “decentralized autonomous artist”, as those who created it describe it, in 2021. They are the “creative” entrepreneur collective ElevenYellow and the German artist and programmer Mario Klingemann. But how does it work? Unlike a classic bot, Botto writes its own prompts, via an algorithm. The creators therefore explained to him what a prompt was, “without any specific aesthetic indication”, explains Simon Hudson, one of the leaders of the project, to CNBC. They also told the bot to use plus or minus signs to emphasize certain aspects. “He started by combining random words, phrases and symbols…to produce images,” recalls Hudson.

Botto thus generates 70,000 images per week. He selects 350 which he presents to a committee of 5,000 people called BottoDAO (for “decentralized autonomous organization”). This committee will choose an image which will then be put up for sale on an NFT platform. The bot will then analyze the committee member voting data to create its next images.

15,000 people have already participated in the project, we can read on the site. Everyone can vote for free. But in general people “contribute to the economy”, according to Hudson, and buy tokens – a kind of digital token. With these tokens, they have “points”, or additional voting rights. And half of the revenue from auction sales goes to voters, in proportion to the points placed (regardless of the work selected). The other half is kept to cover the bot's operating costs.

Legitimization?

But it is not only on an NFT platform that we find these works. And this is where this project – which could be considered “niche” or “geek” and which may recall the NFT bubble of 2021, which has since burst – takes on its full seriousness. In October, the Sotheby's auction house held a sale. Six works sold for a total of $351,000. The whole point is that Sotheby's accepted payments in cryptocurrencies for this auction.

CNBC recalls that these same works had been auctioned with a base price of 13,000 to 15,000 dollars on the NFT market. But they had not found a buyer. “Art” made by an AI therefore finds new clients, wealthier ones at that, outside its basic market. Which remains unknown to the general public.

This “artistic” creation also wants to create legitimacy. This is also one of the project’s missions. “If there's any kind of goal for Botto, it's first to be recognized as an artist, and I think the second is to become a successful artist,” Hudson details. “An artist's success can be looked at from different angles: commercial success, financial success, cultural success, spiritual success – whether he really has a deep impact on people,” he reflects.

It is also with the aspect of community that the project hopes to find its success, change the perception of what art is and stand out in a context where images generated by AI are becoming omnipresent. “With Botto, we get rid of the myth of the solitary brilliant artist and we show that the work of art is in reality a collective process of creating meaning. And with the deluge of AI-generated content, this process will become even more important,” Hudson continues.

Copyright?

Beyond the visual aspect and the debate on the possibility of considering these image productions as “art”, another question arises: that of the legal framework. Who owns this work? An AI or software does not have legal personality and therefore cannot, in theory, claim the status of creator. As for the humans behind AI, their creation is legally limited to computer code. However, these rules date from a time when AI was less efficient and used much less for content creation. With the current proliferation of images, music and other productions generated by AI, the question of adapting legislation becomes crucial. The income generated by initiatives like Botto also fuels this debate.

Another important point concerns the protection of copyright on existing human works. To train AI, developers use an immense amount of tables, images, photos, poems, songs, press articles, and many other works. According to some observers, this practice could constitute a violation of copyright.