In the past 10 years, there has been a great revolution in technologies applicable to the fine art world (e.g. related to custody, conservation and monitoring of artworks). Consequently, the world of collections has been forced to face its technological backwardness.
The two main areas where technology is currently present and applied are:
- Front-end side: in-person and digital interaction with visitors (i.e. electronic ticketing, interaction design etc.). The pandemic has led in many cases to rethinking and restructuring these activities.
- Back-end side: the most interesting side. Approximately 90% of an institutional collector’s activities is neither visible nor known to the public. For instance, some museums have constituted Collection Centers with joint functionalities for study and conservation purposes, leveraging on technologies such as tagging. The most interesting innovations tied in this field are mainly related to internal processes.
Customer journey digitization
Whether it is a private or a public collection, there are always two sides in managing the visitor’s journey: the direct visit experience on the visitor’s side, and its management on the cultural institution’s one.
Unfortunately, the digital maturity assessment conducted on Italian cultural institutions yields to this day rather disappointing results. The digital divide of Italian museums still places them behind their foreign equivalents.
According to a study conducted by ISTAT on a sample of 4800 Italian cultural institutions in 2020:
- 83% does not provide audio/videoguides
- 88% does not have apps for smartphones and tablets
- 86% does not have an online ticketing service
- 75% does not have in place analytics systems
The impact of technologies on the art market
The field of valuations and appraisals is one of the most impacted by technology. The potential of online sales is already known. For instance, in 2020, 25% of auctions took place online compared to 9 % in 2019. Through these selling channels, over $12.4 billions worth of art were transacted, doubling the results of the previous year. Moreover, 90% of collectors with a high disposable income claim to have visited at least one art fair or galleria via digital means in the past year.
The panorama of technological tools currently operating the art world is progressively evolving but at present, such tools are mostly deployed as facilitators of the purchasing and data gathering process, thereby lacking of a decisive role in the approach that collectors have with the art world as a whole.
Amongst emerging players in this field, there are certainly marketplaces (e.g. Artsy), data aggregators (e.g. Artnet) and digital advisors (e.g. Ocula). In the past three years, several B2B and B2C art valuation software have appeared on the market, each with different maturity stages, purposes and data sources. These software provide estimates of the value of single artworks or collections based on the aggregation and analysis of digitally acquired data (e.g. pictures of the artwork, size, catalogue entries etc.). While the system carries out a “pre-screening” phase and accelerates it, human touch and expertise remain crucial for the process’ completion.
Such software match multiple valuation indexes depending on client needs and the initial business model (e.e. art work’s insurance, lending etc.). One of this approach’s weaknesses is that the values are attributed based on results registered on databases of reference, which in the majority of instances only include auction results. The latter are only representative of a portion of the art market, which is undoubtedly important but not the only one. Currently, there are still no univocal estimates on the latter’s weight on the art market as whole. In auction transactions, the seller is known while the buyer often is not and the valuations generated by these software exclude results from informal transactions or those that occur via alternative channels, thus leading to inaccuracies in their estimates.
The role covered by these software is to integrate, support and accelerate decisional process in a transparent way. Nonetheless, perplexities remain pertaining their predictive capacities, especially in instances where the sample is significantly enlarged. When the artwork subject to appraisal is by a lesser known or emerging artist with limited free float and a heterogeneous artistic production, a predictive software’s capacity to fill in the missing data is still to be defined and improved.
The other important aspect to be considered amongst the applications of artificial intelligence to artwork appraisals is related to the quality and accuracy of recommender systems. The latter constitute the main pillar of all platforms such as Amazon or Netflix. Algorithms are at the base of digital services, due to their ability of tracking user behavior, comparing it with that of other users, learning what user preferences are and based on these to generate recommendations of progressively increasing accuracy. Each recommender system functions in a unique way, thanks to its capacity of stimulating diversity, reducing user decision-making workload, increasing user satisfaction and supporting users to define their taste.
In the art world, recommender systems are also being used to support collectors or aspiring ones in their purchase decisions. Currently, only startups have worked towards the development of recommender systems in the artistic sector. However, these often lack of the necessary transaction and client data which are fundamental to progressively improve the recommender system’s accuracy and consequent effectiveness. Recommender systems’ bottleneck is that their predictive capacity is very effective when working on millions of users. When working on limited client and transaction datasets instead, the system’s predictive capacity requires more time to be refined.
Due to this reason, several startups to this day claim to be able of providing an efficient prediction service but they frequently lack of a sufficiently large statistical base for it to be truly effective since it takes time for machine learning to successfully work.
In the future, these recommender systems may conquer a market portion by focusing on a target audience with limited time to research art or who invests in art guided by speculative or other types of purposes. Lastly, according to a study conducted on Italian collectors by Banca Intesa, 93% of collectors claimed to be guided in the purchase of art by the ludic, intellectual or research pleasure. The true driver in their case is the pleasure of discovering something new. The risk for such a pleasure to be reduced or mitigated by the aforementioned recommender systems, inevitably leads one to be skeptical about the possibility for a mass dissemination of these software, especially in the B2C sphere.
Guido Guerzoni is a lecturer of the Museum Management course at Bocconi University in Milan and professor at SDA Bocconi School of Management. He is part of the MIC’s Superior Council and technical Art Advisor of Banca d’Italia’s Museum and of Gallerie d’Intesa San Paolo. He is one of the most expert economists of culture.