Bloomberg: How AI and Real Time Valuation help bring transparency to the art market


Our CEO and co-founder, Rob Steinberg, recently sat down with Bloomberg Radio to talk about how ARTBnk is using small data and artificial intelligence to provide Real Time Valuations (RTV) for works of art and the challenges of getting there.

“Everyone is talking about big data and AI, and there isn’t big data with art. We figured out how to use small data and we’re doing that very successfully.”

If you’re not familiar with the concept of RTV, think of it like Zillow's Zestimate for homes, and Kelley's Blue Book for vehicles. RTVs are recalculated and adjusted daily based on market changes providing you with the most current valuation available.

Turn up your sound and enjoy the next 8 minutes or check out the full transcript below.


Pat Carroll: Coming up what is your art worth? There is an algo for that now. We will have the CEO of ArtBnk which is creating formulas for you to instantly know the real value of your artwork.

Tom Maroney: Joining us in studio Rob Steinberg the CEO of ArtBnk. A-R-T-B-N-K a missing ‘A’ just for you spellers out there. He wants us to think Zillow’s Zestimate for homes or Kelley Blue Book for vehicles. It’s a mobile app. It’s now being tested by 500 Beta testers and Rob you are telling me this would be out the end of May for general consumption. First of all, welcome to the program tell us all about this new gismo.

Rob Steinberg: Thank you. Thanks for having me. In the simplest terms it is really a fintech tool. Its mark to market forward. Hasn’t existed before. In greater detail it’s a platform it has a collection management system and it also delivers instant, objective, reliable, and accurate valuation of what art is worth today.

TM: So, tell us, walk us through how someone would use this.

RS: Go on the website, sign up, its got a desktop app as well as mobile app. Very simple interface. Either you drag and drop from your desktop or upload. If the object is in our database it populates the necessary information about the artist and the object. It might give you suggestions if it doesn’t know there is a lot of image recognition then you can add some other details and if it is an object not necessarily one thats been sold in our lifetime but it has a secondary market we will give you what it’s worth today.

PC: Well, let’s talk about this secondary market. You can’t actually appraise all art out there.

RS: Absolutely. And for now, you know we are a young company, so we got approximately 150 artists generating real time valuation; there will be 200 by the time we launch, 500 by the end of the year…that’s a surprisingly large percentage of the art market. Primary artists who only work through first sales is an entirely a different realm, and we’ll get there. We are working with insurance companies, for example. We will start accumulating private sale data, but right now is a little bit of liar’s poker, so we can’t use private sale data; we are entirely dependent on what the public record shows.

TM: Our guest Rob Steinberg, CEO of ARTBNK, a mobile app that helps you…well give you a valuation in real time of a piece of art dealing mostly as Rob explains in the secondary market. You know, Rob, when I first came upon this idea of yours, I thought to myself why hasn’t been done long before this? What’s the answer?

RS: Its really simple it couldn’t have been done.

TM: And why?

RS: The technology that we’re using wasn’t available and we are doing a couple of really important things with it. We’re using a lot of artificial intelligence and neural networks and machine learning and image recognition, but in a breakthrough way. The major simplest explanation—for the last year or two everybody has been talking about it—big data. And AI’s are dumb and they crunch through petabytes of data and get trained…they are a little bit like a golden retriever, you gotta train them to do what they need to do. But that’s of limited use because art is a great example—there isn’t big data. Now the buzz is: ‘oh, we need to figure out how to use small data with AI’s’ and we’re doing that very successfully and we think that is a breakthrough.

PC: So you say you go over 6000 data points—

RS: Exactly.

PC: What else can you assess besides previous sale price? What else are you looking at?

RS: Oh, if you look at substrate, medium, price moves…one of the things that I think is most important is trends. So, there are two things that are important to give an accurate value in a stable market but also predict what would move art. So there are the macrotrends—stock market, interest rates, a dealer who picks ups an artists or drops an artist can affect the market value, a major show or retrospective—the list goes on and on and on and they have various ways according to what else is in the equation.

PC: So you’re trying to predict the desires and the personality of the rich in a way—

RS: (laughs)

PC: —what gets their attention or what gets other people’s attention that gets their attention.

RS: Well they’re leading indicators. You can never predict what’ll happen on any given day. There are examples from last year a Basquiat painting sold for USD$110 million and it was a price no object…get-out-of-my-way-I’m-buying-it-at-any-price and there are other reasons other than its value to pay that money. Basquiat’s work was not elevated by that sale. As a matter of fact, so much work came onto the market that it depressed prices for a while. And none of it was worth $110 and after a while it stabilized. So that sort of event you look at the Da Vinci sale from last year…I think the 400 or some million-dollar sale was a third of the Gross National Product of St. Kitts. Is a painting worth that? The joke there was…

PC: In that shape too, we should add. In that condition for that painting.

RS: …The joke there is that it was in the contemporary sale because most of it had been painted in the last decade.

TM: You know Rob it seems to me too that people are going to use this as a price check. So, when they’re out and they’re in the midst of a transaction, they’re gonna say ‘Well I’m gonna check my mobile app for the sum.’ So I’m not sure dealers are gonna be crazy about this technology, right?

RS: Well, it’s a great point and some…and I think the preponderance of dealers will be threatened…but what happens in any market that becomes transparent and you change from asymmetrical information owned by the experts—a guild really—when it’s more publicly available, a few things happen: absolute volume goes up, margins tend to go down, but there’s a great increase in the amount of profits in the industry. You see it in Wall Street: some decades ago, trading volume was in the millions of shares a day and it cost a dollar a share to trade it and now it’s billions of shares a day for fractions of a cent a share; the flow of information did that. And some dealers will do very well. Millennials are coming into the market, they need confidence, they need common information, because it is very easy the risk of being foolish about art. So that transparency I think will help dealers that understand it and dealers that don’t will be in trouble.

JW: Rob Steinberg, CEO of ARTBNK, you talk about standardizing art valuation. Is there sort of a rubric? What are the standards? How is that working?

RS: Standards in art…wow. (laughs).

JW: Well for your purposes, for what you’re trying to do.

RS: So its actually the other breakthrough that we had to do and it was a bad surprise. When we realized that the databases that are public that everybody uses—and there are millions of records and lots of sources to access those records—but in terms of modern technology for AI they’re useless. There are two examples I can give you. One is Picasso: you and I know there is one of them, in the databases there are fifteen. There was one Christie’s auction in three lots—same auction—there were three different Picassos as far as computers concerned database, concerned different life dates, death dates, country of death, country of birth, that had to be cleaned up. More important example is a great artist named Gerhard Richter who uses a substrate called aludo bond—that’s actually aluminum sheet. In Great Britain that would be aluminium, here it’s aluminum; there are about fifteen brands which may or may not affect the value of the work, but we had to go into the database and clean all that up and automate it.

TM: Alright, well Rob Steinberg we’re gonna have to limit the discussion here. Fascinating. This mobile app gives you what Rob explains as real time valuation of artwork; will become open to the general public at the end of May.

RS: And you can sign up now. It’s

TM: And they are based in Newmarket, New Hampshire. Thanks for stopping by.

RS: Thanks so much.

Seeing is Believing

See for yourself how ARTBnk can help you organize your art collection, alert you to upcoming auction sales, and provide a current valuation for most artwork sold in the secondary market.

Get started with ARTBnk today!