Is there a danger that when autocomplete steps in while we type, it may be interfering with the task we are trying to perform?

Iterations of a Generative Adversarial Network AI learning to create abstract art

Pop a few letters into the text box of your chosen search engine and it’s likely that, before you finish typing, a handful of suggested phrases will have appeared. These autocomplete choices are based on millions of queries inputted by other users before you. They can seem helpful, saving you a few seconds of typing or even offering alternative ways of phrasing your query.

We are used to this sort of thing happening in our offline lives all the time. During a successful conversation with someone, it is common to complete each other's sentences. This behaviour can be a sign we are tuned into each one another’s thoughts.

But this process is not always so productive – our brains fetch the words we use through a complicated set of steps that involve sifting through a pile of other associated words. If we are presented with an unrelated set of words during this process, it can make that retrieval process take longer – an effect known in psycho-linguistics as semantic interference.

Is there a danger that when autocomplete steps in while we type, it may be interfering with the task we are trying to perform? Could it even lead you to choose words you might otherwise not have? There is very little research currently on the effect autocomplete has on us and our ability to find answers effectively, but these are questions we should be asking.

It is also just one example of how our search for information online can become polluted. In this case it is the search activity of other people that leads to suggestions that can alter what we find. The architecture of search engines themselves can also affect the result they produce.

But what if you wanted to find pages from another country or on pages not considered to be as readable? Some of this can be overcome by tinkering with your search settings, trying different key words or going far back through the results to find what you need, but not all. And few people have the time or patience to keep doing this.

Many search engines also now rank sponsored pages – those that the owner has paid the search engine provider to promote – high on the first page of results. If we are in a hurry, it can be all too easy to click one of these rather than delve deeper into the search results.

Our own previous search activity can also affect the results we are given. We rarely use the internet for a single task. I might, for example, be looking for a wheelbarrow I need in my garden, and then doing some academic research later in the day. But that earlier search history can stick with you, and this can bias the results you are served up later. This is problematic, for example, if I was searching for two different, divergent viewpoints on the same topic in an attempt to perhaps understand different communities on either side of an argument. It could only serve to exacerbate the bubbles we find ourselves in rather than allowing us to access information on all sides of an argument.

The search engines themselves use all this to help reduce the time it takes to retrieve search results, and so, make it quicker for users to find information. But what is less clear is whether they might also be interfering with the way we access information and so ultimately affecting our productivity? There is some evidence, for example, that the more we use search engines, the more dependent we become on using them too. Should we be worried about whether this reliance affects our ability to find out information in other ways?

There are, of course, ways around some of these issues. There are search tools that are designed to constrain their results to a specific task we are trying to achieve – searching for academic papers, for example, looking for flights or shopping. There are also search engines that do not use our search history and offer more manual control over what we search for. They are, inevitably, more time consuming to use, but there is also less interference in the task we might be trying to perform.

But giving people too many incomprehensible options and controls to determine what information they are being served with can come with drawbacks. There is a point where the cognitive load needed to get what we want becomes almost too much. My colleagues and I ran some research a few years ago to look at whether people understood social media privacy settings at the time – things like whether they can determine who sees a message or not – and most people couldn’t complete the task.

Perhaps a wider issue is at stake here. Many of the internet platforms we now use to access information, whether those are on search or social media, lie within walled gardens. There is little or no interoperability, as the Furman review on digital competition for the UK Government concluded.

Most other forms of communication we use today are open services – there are shared protocols that allow you to phone anyone else with a phone, for example, or send an email to other email systems. Social media, and in particular instant messaging, doesn’t work this way. We are forced to only get information in a specific way from a limited number of sources. Search engines also offer us little way of comparing the results each of them serve up.

This is strangling innovation in how information is shared and presented to us. The world is not a homogenous place - there are cultural differences and nuanced ways of communicating. I was involved in a project to help build an online genocide memorial in Rwanda. We were completely out of our depth in terms of understanding the nuances of the discussions that were going on, and the only way it could work was by using the expertise of people from the communities themselves.

As we try to navigate the ever growing repositories of information that are available at our fingertips, we are going to need far more innovation, at a local level, to help all of us find what is important to us. 

Further reading

This blog is one of a series of perspective pieces published to support the Royal Society's Online information environment report, which provides an overview of how the internet has changed, and continues to change, the way society engages with scientific information, and how it may be affecting people’s decision-making behaviour.

Authors

  • Professor Derek McAuley

    Professor Derek McAuley