FRIEDRICH-EBERT-STIFTUNG – Artificial Intelligence and Auto­mation in Retail ised offers, now smartphone scan and go payment systems, wi-fi and/or Bluetooth-enabled sensors and cameras are increasingly being synchronised to profile consumers’ habits and firms can use such multiple data points to construct ›heat-maps‹ of stores and customer profiles, monitoring flow and lingering. Online these systems can also inform dynamic pricing that can be used to entice customers to buy. The potential benefits of these systems for consumers and retailers are that customers can receive updates and offers tailored to their interests and preferences, increasing the chances that consumers being able to find the products that they want and increasing sales. This can also indirectly benefit workers where increased sales is accompanied by greater demand for labour and job security. However, there are also some potential risks and challenges. For consumers, there is a risk of overspending and overconsumption, par­ticularly if the temptation of tailored offerings is combined with easy access to credit and round the clock shopping. Also, widespread data collection and sharing raises concerns about transparency and privacy. Many consumers are sim­ply unaware of the level of data-gathering by firms in both e-c­ommerce and in-store or the types of data collected. Spanish supermarket chain Mercadona was recently fined EUR 2.5m by regulators under GDPR legislation due to its processing of facial recognition data for all customers, os­tensibly to track shoplifters. 27 The implications of these technologies for skills and careers are not easy to predict. On the one hand, these systems mirror traditional techniques of upselling and cross-selling, typically a skilled task performed by a sales worker engaged in selling high-value items. However, predictive marketing systems can be applied to products across the price range due to the reduced cost of suggestion algorithms and can be made at a distance to the physical store. Thus, in some ways these systems replace the traditional role of retail workers to make personalised recommendations to customers, poten­tially undermining sales workers’ professional identity. On the other hand, algorithmic customer recommendations can function to enhance skills for workers. If algorithms produce a range of possibilities, the sales worker can talk a customer through the various options on offer to help them make an informed decision. This requires investment in sales workers’ product knowledge and digital skills. existing infrastructure, but doing so allows Amazon to collect data on what does and does not sell enabling them to de­velop and market equivalent offerings in direct competition with sellers. There were also concerns that online platforms also have the advantage of being able to profit directly from being able to sell anonymised data to third parties. 2.8 AUTONOMOUS WAREHOUSES, VEHICLES AND ROBOTICS While robots and robotic processes have been in use for some time, particularly in the manufacturing sector, ad­vances in robotics and machine learning have greatly ex­panded the uses that robots and ›intelligent‹ machines can be put to. In the retail sector this includes automated sorting, fault detection and quality checking systems, autonomous vehicles and warehouse robots, as well as a robotic process automation(RPA) systems that can perform various customer service, marketing and HR functions. As with many automa­tion technologies of the past, while such systems automate many physically demanding or repetitive tasks – creating productivity gains and potentially making jobs less arduous or monotonous – there introduction is not without issues. One characteristic of robots and automated computer sys­tems is that they do not get tired and in some cases can perform tasks faster than human workers. This can mean productivity gains for employers and cheaper prices for con­sumers. However, robots are not cheap and sophisticated RPA systems require large amounts of data to develop. This means that for some uses introduction only becomes cost-ef­fective at scale and many smaller retailers may be priced out of making the most of these technologies or may have to pay for third party providers for some services, both of which have implications for their ability to compete on price. Fur­ther, the benefits of deploying such technologies at scale will likely favour larger players, prompting further consolidation of the industry. For workers, the balance of potential impacts are unclear. On the one hand, where robots are used to automate physically demanding or repetitive tasks this can improve health and safety and free up workers’ time to do more specialist tasks leading to job upskilling. On the other hand, there are a number of concerns related to jobs and health and safety. Given the opportunities for data collection available online, combined with existing technical capabilities, there were concerns among focus group participants that predictive marketing systems give larger retailers with an existing on­line presence an advantage over smaller bricks-and-mortar retailers. While it was noted that some smaller retailers are moving in this space, and that some were making use of innovative individual and collective store loyalty schemes to do so, there were concerns about the ownership and sharing of data. There were questions about who owned and had access to data, particularly where smaller retailers used large online platforms to sell their products. As noted in section 2.1, some retailers are turning to platforms such as Amazon Marketplace to sell their products and make use of their In relation to jobs there are three main concerns. First, that automation may lead to less demand for human workers and fewer jobs. Second, that removal of repetitive and less-com­plex tasks may be something of a double-edged sword. The removal of time-consuming or mundane tasks frees up time for workers to do more stimulating tasks closer to the customer, opening possibilities for upskilling and career de­velopment. However, this could lead to less diversification of tasks, leaving only complex and demanding tasks as workers cannot switch between easier and more taxing tasks as part of their work day, leading to an intensification of work. A related, third, concern is that any job upskilling needs to be accompanied by: i) the appropriate training, and ii) increased pay commensurate with the work. There were concerns 10