Latest news about Bitcoin and all cryptocurrencies. Your daily crypto news habit.
Chainalysis has reported that the NFT market grew to $40.9 billion in 2021, having barely reached $1 billion in 2020. Now the time has come for artificial intelligence to tap into this mysterious domain in search of big profit. There are starting to appear projects that create AI-generative NFTs for auction, some of them do it with human participation, some solely rely on artificial intelligence.
Projects that create AI-generative NFTs
Here I will cover some of the standout ones and describe their work principles. The basic technological element of these systems is a GAN (generative adversarial network). It consists of a pair of neural networks one of which generates artworks and the other called discriminator selects the ones that it deems valuable from the artistic point of view. The discriminator bases its selection on the recurrence of patterns identified in the images downloaded into the neural network. Sometimes the role of the discriminator is shared by a human community.
Botto has so far been one of the most conspicuous examples of AI-generative NFTs by producing six artworks that have been cumulatively sold for $1.3 million. Botto produces several hundred artworks per day and a community of real people select the artworks that will go to auction.
In creating its artworks, Botto starts with a string of words that it feeds into VQGAN (Visual Quantized Generative Adversarial Network). VQGAN is a neural network that produces images based on these word strings and a database of artworks fed into it. Once it completes an artwork, it feeds it to the discriminator CLIP (Contrastive Language-Image Pre-training) that cross-references the initial word string with the image produced and determines what adjustments need to be made by VQGAN so that the image meets certain criteria.
Once a final image is completed, CLIP generates a two-word title for it. Afterwards, a description needs to be done; for this Botto recruits the neural network GPT-3 of OpenAI to generate a description for the artwork. GPT-3 comes up with 5–10 poetry descriptions of which one is picked and edited by Botto’s core human team.
Botto generates 300 artworks a day and 350 of them are chosen each week by a ‘taste model’ for the final verdict of Botto’s community of thousands of people. There is only one winner each week, and it gets minted as an NFT and is then auctioned on SuperRare.
Even before NFTs, AI-generative art was going for pricy sums. Thus, in October 2018, a group of French artists called Obvious sold The Portrait of Edward de Belamy at an auction at Christie’s for $432,000. The portrait was created by a GAN using a database of 15,000 portraits painted between the 14th and 20th centuries.
How will it affect human artists?
I am more of the belief that AI-generative art will not threaten the appreciation of artworks painted by professional human artists. For the mere fact that collectors always appreciate a painter’s genius materialized in their art.
Another drastic argument in the favor of human artists is the fact that AI-generative art is only present in the digital form and cannot be recreated by artificial intelligence in the physical form. Only an artist’s hand could make a digital painting into a physical copy if such need arises.
Painters could actually benefit from the use of artificial intelligence and neural networks to find inspiration for their own creations and probably even mention some of the distinctive AI’s as contributors to their art as a source of ideas to raise the price of their works.
And artificial intelligence can become a gateway for many people who are not professional painters but have different talents to visualize their ideas in highly artistic forms. This can equally be a fantastic opportunity for poets to create visual artworks based on their poetry and can also find many use cases for the general public when a certain verbalized idea needs to be represented as a visual.
In the end, many things have gone through industrialisation and digitisation, but people have not suffered from the technological advances achieved by civilisation. And with visual art, many things have been modernised over the last two decades: digital art has become an everyday reality, and painters have mastered digital tools to make their creations. And I do not see why any harm should be inflicted through another step forward.
I would not be worried about an AI takeover of the world of art. I would rather be looking for positives and use cases that artworks generated by artificial intelligence can offer to people. And the NFT medium will only help to bridge the bureaucratic gap between the seller and the buyer of NFT art. And the NFT’s original address will likewise help understand if the artwork was created by an artificial intelligence. If the contract that minted the NFT belongs to an AI project, the public will be able to see it.
Author Bio
Galina Likhitskaya is the Vice President, Operation & Product of a smart contract audit company HashEx. She has profound experience in business development, project management, development of B2C & B2B products, marketing for IT and fintech projects. As the Vice President of HashEx, Galina is primarily responsible for building operational processes, developing company products, defining business goals and planning the company's activities in accordance with its long-term mission. Â
Â
Image: Edmond de Belamy (Wikipedia)
Disclaimer
The views and opinions expressed in this article are solely those of the authors and do not reflect the views of Bitcoin Insider. Every investment and trading move involves risk - this is especially true for cryptocurrencies given their volatility. We strongly advise our readers to conduct their own research when making a decision.