The internet has opened a huge window onto the range of products and services available to consumers, and revolutionised the way we purchase them. In the UK alone, E-commerce sales have rocketed up fifty-six per cent from £375.1 billion in 2009 to £586 billion in 2017. In some cases, the choice of products is so vast that to compare them as an individual consumer would be nigh on impossible, especially in complex fields such as energy and insurance products. One of the consequences of this is the growth of what are known as algorithmic consumers.
Jesse Norman MP, in his essay in Britain Beyond Brexit, is right to point out that an asymmetry of information has opened up between companies with access to vast quantities of data and expertise to produce complex products, and the consumers who purchase them. Algorithmic consumers – essentially software that can find a product to fit a consumer’s specific needs – can assist in re-balancing the equation.
An algorithm is a decision-making process that employs a set of rules and procedures to supply outcomes based on data inputs. People use similar decision-making processes in their daily lives when, for example, deciding when and what to eat. They will determine how hungry they are, what food they have available, how tasty or healthy each option is, and then they will weigh the answers to these questions (or inputs) in order to reach an outcome in accordance with their preferences.
At present, many algorithms function as tools to increase choice and to help consumers make better decisions. Simple algorithms collect and organise information relating to flights, hotels, car hire; more sophisticated ones, like those used by online dating services, use consumers’ characteristics and past preferences to narrow down options and present outcomes assumed to be more desirable.
The algorithmic consumers of the near-future will be able to talk to each other, identify a need, search for an optimal product or service and execute a transaction. Applied to, say, switching energy supplier, an algorithmic consumer would increase the consumer’s power: weighing up the advantages of different energy tariffs and contracts and deciding on the best and cheapest deal. This would make easier a task that is so complicated and time consuming that most people chose to not attempt it, even if it in their interest to do so.
Algorithmic consumers, however, come with their own risks and potential harms. These systems could be said to reduce autonomy as consumers are distanced from making their own choices from an open field. Algorithmic consumers could be overly deferential to precedent – for example an algorithm that purchased food cupboard essentials could assume that what you ate last week implies that you would want to eat the same again in successive weeks. And then there is the danger that some systems could become so sophisticated that even their developers do not completely understand the basis on which they make purchasing decisions.
These risks could be mitigated by requiring algorithmic consumer systems to ensure their users are aware of the parameters by which they make decisions, and allowing them to be changed. Competition between providers of these systems – and lowering barriers to entry into the sector – could be encouraged, thereby providing consumers with a choice of consumer algorithms: so the consumer annoyed at the delivery of the same basket of groceries for the 6th week running can switch to another system that doesn’t assume that she’d want to eat the same thing every week.