AI replace human decision-making

February 4th, 2021

Myths Busted: AI is no Replacement for Human Decision Making

Artificial intelligence (AI) is now firmly within the mainstream. Microsoft recently found that 56% of UK organisations are using it to some degree. The same research also pointed to a clear competitive advantage linked to AI, with businesses already using it performing an average of 11.5% better than those who are not.

For forward-thinking businesses, technologies such as machine learning, national language processing and predictive modelling are helping them make sense of potentially vast amounts of data, reduce error and boost output.

But what happens when the machines go one step further? It is one thing for algorithmic analysis to flag up a problem. The controversy arises where the system automatically generates a solution, decides on a course of action and executes it, eliminating the need for human intervention.

It is easy to see why individual employees may push back against this type of technology, and how this could be a very real barrier to transformation. After all, why would you actively welcome a new tool, if it threatens to make your role redundant?

Meanwhile, last summer’s exam furore over predicted grades demonstrates that algorithms don’t always get it right. Taking into account issues such as accountability, regulatory oversight and company reputation, businesses themselves are right to be wary of relying solely on AI for important decisions.

So how do you get it right? As we’ll see, the most effective uses of AI within the workplace don’t actually replace human decision making. Rather, they enhance it. Here’s how…

AI frees up bandwidth

What do we want from our departmental managers, finance team and other key staff?

Almost certainty, if you can possibly help it, you do not want highly-skilled employees bogged down in routine, transactional tasks. You want them to put their expertise directly to work, solving business problems and driving strategy.

This actually dovetails with what employees themselves want. Direct involvement in the decision-making process tends to boost engagement. And as Gallup demonstrated, highly engaged employees tend to produce better outcomes.

So where does AI fit into this? The fear is sometimes that the software will end up doing the decision-making for you, resulting in reduced scope for human input. In reality, most businesses find that the reverse is true.

AI lets employees process and analyse data much faster and more accurately than they would otherwise. Take your accounts department, for instance: through machine learning, they have the potential to process transactions, to automatically unearth and address irregularities in record time, and with the minimum of human intervention. The time and input required for routine reporting is dramatically reduced.

Meanwhile, with solutions linked not just to finance but also to areas such as manufacturing, logistics, marketing and customer care, AI concepts such as natural language processing are being put to work. For example, it’s becoming possible for employees to execute all manner of routine tasks simply by asking a voice-enabled communications assistant.

Businesses are increasingly finding that AI is reducing the time required for necessary but routine work. These are precisely the type of tasks that eat into employees’ time and prevent them from taking a more active role in decision making.

PwC found that in forward-thinking firms, 75% of business analysts’ time is spent on developing insight. In simple terms, if you want humans to bring their experience to the table and become more active in decision making, AI is pretty much essential technology.

AI delivers the full picture

What do customers really think about our brand? Where is the next big trend coming from? How can we tell when a client is about to leave for a competitor, or a piece of machinery is about to malfunction?

Conventional performance management solutions and other types of business software are fine for basic number crunching but isn’t always capable of answering these kinds of big questions.

Humans are much better at interpreting nuance. Trouble is, we cannot be expected to read absolutely everything that may be relevant to business decisions, and we can’t be on call 24/7.

An estimated 73% of company data goes unused for data analysis. Often, this is because certain datasets are too difficult to interpret: data is unstructured, or else there’s just too much of it.

AI happens to be extremely useful at sifting through and making sense of data that would otherwise be out of bounds to decision makers. Examples include sentiment analysis tools that can constantly scan swathes of content on social media platforms to pick out insights, or finance regulatory tools that can ‘read’ complex documents and flag up compliance issues. Over time, a manufacturing plant monitoring tool can ‘learn’ to recognise the various combinations of readings that might indicate performance issues. When these circumstances arise, the issue is automatically flagged up.

In these use cases, AI opens up data streams that might otherwise be difficult to interpret in large quantities. It doesn’t have to make the role of human decision makers redundant. Rather, it picks up on insights that would otherwise be missed and makes sure you are in possession of the full facts before deciding what action to take.

Looking to the future

What about AI’s role when it comes to day-to-day processing decisions?

Take decisions linked to consumer credit, for instance. Where individual employees are left to decide what credit options should be made available to customers, it’s easy for bias to creep in, or for inconsistencies to emerge. If you have an AI-based tool that’s able to process a customer’s details and assess their risk based on set rules, there’s the potential for much more consistent decision making across the business.

But of course, any AI solution is only as effective as the algorithm behind it. To avoid bias (along with risky approvals), continued human oversight is essential. Otherwise, you risk systemising into the decision-making process the very things you want to avoid.

At its best, AI disrupts the human decision-making process. With the ability to read and interpret vast quantities of data, it has the potential to put relevant information at your fingertips faster than ever before.

A recent estimate for the banking sector suggests that we can soon expect decision-making processes to be 34% informed by machine algorithms and 66% by human judgment. This is probably a realistic interpretation of what the future holds: AI will have a big part to play in informing decisions, but the final decision will remain with people.