COVID-19: Our deliveries are guaranteed and our customer service is open!
Find out more
When he grew up in West Africa, SuperFluid Labs CEO Timothy Kotin represented Ghana at the International Physics Olympiad. On other occasions, Kotin would compete in global tournaments in mathematics. Some years later, the prodigy crossed the Atlantic to pick up an undergraduate degree in computer science and electrical engineering at Harvard University in Cambridge, Massachusetts. An MPhil at the other Cambridge in the UK soon followed, before Kotin went back to Africa to use artificial intelligence (AI) to unlock the power of data.
Having founded Nairobi-based pioneering analytics firm SuperFluid Labs in 2015, Kotin has now teamed up with Art-tech platform Turbare in 2020, to deploy AI and build a machine learning model that it says can accurately predict the true value of art.
How this revolutionary team builds their machine model is both complex and distinctly logical. More than 12,000 pieces of art from Africa, which had previously sold at various sales and auctions have been fed into a specially designed computer programme.
Numerous attributes and characteristics for every single piece of art is pumped into an algorithm. Everything from the dimension of the piece, the theme, the location, the country of origin of the artist, their age, and even external data sources like an artists social media following and their activity, is registered.
The machine learning model is then able to use this data to provide a ‘fair market value’ estimate on the turbare.com website, a starting point that collectors and art lovers use to make a bid on a particular piece of art. Future market trends are also factored into forecast over 1, 3 and 5 year intervals.
“Now, the way the machine works is that the model learned the characteristics of different pieces of art, and to what extent each of those attributes contributed to the final price” says Mr Kotin.
“Once the model had been able to learn these pieces, and the kind of contributions that each of these characteristics made to the final price, the model was then tested against new pieces of art that it had not been exposed to previously, and it was challenged to make a prediction on what those prices would be. This was an intuitive process, after which the model became highly tuned and sensitive at predicting what characteristics, or what combination of attributes, really contribute to the final price range which a piece of art might be sold. We were able to get the initial versions of this model to an accuracy of 90%.”
For SuperFluid, an analytics revolution in the market is valuable on three different levels.
Accurate machine learning models solves the industry’s over reliance on subjective decision making around the value of artwork. Without AI, it’s almost impossible to do this at scale in any coherent way, says Kotin, unless you rely on individual human experts and their own perspectives. For the burgeoning African art market, showcasing the works of artists from 54 countries, coming from the world’s most diverse continent, AI provides efficiency and greater inclusivity at scale.
At their core, artificial intelligence models can be trained on historical data sets, to learn and perform valuation tasks, as well as human beings can. This means AI gives art collectors greater transparency for buying online, a higher degree of certainty in their investment, and less subjectivity in the market.
It also provides an opportunity for emerging artists who haven’t had the necessary exposure, to quickly receive valuations on their artwork. How their pieces compare with other artists can also be computed by the model, based on similar measurable attributes. Turbare says this will provide further opportunities for emerging artists from the continent to showcase their talent, and earn a living off their creativity.
For galleries, Turbare is able to leverage their model to provide indicative prices with a high degree of accuracy. This is valuable to galleries that are looking to sell pieces listed on the platform, says Kotin, and are “looking to get an indication of what the consumer appetite or demand for those particular pieces are.”
The machine model isn’t yet perfect, but as new data is added, it will compel the model to improve. As customer bases grow, and individuals vote with their bids, secondary data streams will further improve accuracy.
How many articles have been written about a particular artist can affect valuation, as can macroeconomic environments and seasonal effects. These signals impact consumer demand for art, and influence how much people are willing to pay for it. “Over time, this should be a third way in which we expect the models to become even more accurate,” Kotin told Turbare.
Predicting the price of artwork that sells the most at auction remains a challenge however, as until there’s a growing data set at the top end, it’s harder to generalise.
It’s one of the remaining thorny issues for AI models in art, as they can struggle to foretell the emotional pull of a particular work on the pursestrings of an individual buyer in the room – or watching online – as Covid-19 radically alters the way we view and buy art.
Much of the art world has shifted online in 2020, with exhibitions becoming virtual shows, and galleries have been forced to build digital showrooms as margins squeeze. It means that galleries with an online presence have proven more robust in the face of a global pandemic, while artists selling work on the web have fared better than most.
For the international art market, historically opaque in nature, Covid-19 has presented a golden opportunity for radical reinvention.
“There are certain industries that have undoubtedly been set back,” says Mr Kotin. “But there’s also a number of industries that have been pulled forward by almost a decade. Everything from the art industry to e-learning and e-commerce have been accelerated in a way that would have been unimaginable five years ago, or even two years ago. It has become essential for businesses to embrace digital technologies as a way of reaching their customers, and better understanding the evolving consumer sentiment, so that businesses can promptly react. That’s definitely a positive trend.”
For the burgeoning African and Middle Eastern art markets in particular, use of AI has the potential to draw in a new generation of tech-savvy collectors. Investors who are interested in a more diverse group of artists, who want reassurance that the artwork they are investing in will hold value, or better still appreciate over time, can be attracted as African culture goes global. The historical neglect shown to world class artists from the continent and other international markets can therefore help to be addressed.
For the polymath from Ghana, artists from the African continent have the power to inspire on the world stage, as appreciation grows among young professionals wishing to collect.
Art collections used to be something that only the ultra rich would consider after they had discretionary income or savings, Kotin says, but the work being done by the likes of October Gallery and Turbare will open up the market further.
“That is fantastic news for the individual artists, who will have a global stage to feature their work, but also for consumers and individuals, who will be exposed and enchanted by African art.”
Turbare speaks to founders Joe Anka & David Hutchful on how they are disrupting the global art market and creating opportunities for collectors and artists alike.
Turbare speaks to Jason Bailey, founder of Artnome, an analytical digital art database that’s helping improve opportunities for collectors and artists alike.
In America, Artnome are building the world’s largest database of paintings, using Artificial Intelligence and an eye-watering amount of data
Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.
Any cookies that may not be particularly necessary for the website to function and is used specifically to collect user personal data via analytics, ads, other embedded contents are termed as non-necessary cookies. It is mandatory to procure user consent prior to running these cookies on your website.