“AI technology invariably needs human beings”

Artificial intelligence is changing our life and the way we do business. But it also raises strong concerns. Simon Carter, Head of Deutsche Bank’s Data Innovation Group, describes how AI can enhance human work, rather than replacing it.

Video: Why AI is useless without humans

Why has AI become so important in the business world?

Artificial intelligence is changing industries, from healthcare to finance, to media. It can increase productivity and create new revenue streams. And it can rapidly sift through enormous, highly complex sets of data to provide companies with deep insights.

Let’s look at a specific example. How can AI help a retail business manager deciding where to open a new shop?

A business owner can tap into a wide variety of data sources to help make this decision. Examples include mobile network data, census reports and demographic data, and accessibility data for a specific location.

But the real beauty of AI is that it can combine these different sets of data to draw insights for example by looking for locations that are both highly accessible to large numbers of customers and where there are high levels of customer traffic (using mobility data). AI can also derive data from online rating portals such as Yelp to find out how much people like a place.

How can a machine extract human sentiment out of big data?

This is actually an extremely difficult problem to solve, in particular when you have to take into account the fact that human expression is very dependent on context that can change and even reverse the apparent sentiment – whether written or verbal. Artificial intelligence algorithms have to undergo intense training under the guidance of human operators before they can even begin to identify sentiment patterns.

AI can be very effective at analysing very specific questions and doing very specific jobs.

How good are these models today?

Basic sentiment models simply look for keywords that represent a particular sentiment, for example people using the word “happy” in a social media post. More complex algorithms will try to take into account patterns of speech and context, but these are far from perfect.

Ok, but how can you avoid drawing erroneous conclusions, if no human being can check the datasets?

AI can be very effective at analysing very specific questions and doing very specific jobs. Changing circumstances, and unexpected events, can lead to failures in AI algorithms, meaning that human beings will remain a key part of any business’s decision-making process. I like to think of the future being more akin to having a human decision maker assisted by AI technologies, rather than a business being managed solely by AI algorithms with no human intervention.

Speaking of the future, how fast AI is going to disrupt the world as we know it?

Technology evolution in this area is likely to continue at a high pace, across many different areas. But most entrepreneurs are likely to experience a more gradual and progressive evolution, as the winning technologies often take time to gain wide acceptance.

It is very unlikely that AI will be any better than human beings at anticipating future developments or changes in behaviour anytime soon.

Can AI go beyond observing current state of play and anticipate future developments or changes in people’s behaviour?

AI can certainly provide additional insights and analysis to help guide businesses through future economic conditions. It is very unlikely that AI will be any better than human beings at anticipating future developments or changes in behaviour anytime soon – just consider how unreliable weather forecasts are, even when derived from AI algorithms.

Is there a scenario where AI might be a better option than human intelligence?

AI technology invariably needs human beings. It must be developed and trained by people to perform specific, precisely defined tasks. Humans will still be needed to define the questions the AI will be tasked to answer, as well as interpret the output from this technology.

On top of this, people will continue to be essential to execute any strategies developed off the back of AI-derived insights. We are a very, very long way from a world in which artificial intelligence machines run the show.

About Simon Carter

Simon Carter heads Deutsche Bank’s Data Innovation Group, which analyses and transforms big datasets into specific and relevant pieces of information for its clients (corporates, asset managers, research analysts, etc.).

He joined Deutsche Bank in 2010 as Head of European Equity Derivatives Strategy, having previously been Head of Equity Derivatives Research at BNP Paribas. Simon graduated from Cambridge University in 2000.

Yann Couronneaud

… has followed the advance of AI with a mixture of fascination and unease. Films and series have given tremendous airtime to myths and misconceptions about AI. It is important to understand the benefits and risks AI really bring, and how AI will affect our lives.

Recommended content

Digital Disruption | Outlook

How Artificial Intelligence is changing banking How artificial intelligence is changing banking

AI is considered one of the technologies that can fundamentally change industries. Banking is no exception. We show three possibilities.

How Artificial Intelligence is changing banking Three short videos

Digital Disruption | Story

Digital Disruption | Opinion

What Next: our topics

Link to Responsible Growth
Link to Digital Disruption
Link to Entrepreneurial Success