Bytes on the brain: Can artificial intelligence make better decisions in business?

Today, Artificial Intelligence (AI) supports decision-making in sectors as varied as financial services, healthcare, retail and information technology and has brought about innovative partnerships between businesses, organizations and technology startups. AI technologies, such as deep learning and computer vision, are already used to collect and process data, and to detect patterns and predict behavior. AI can understand natural language and apply reasoning to problemsolving and risk calculation.

Blue Yonder, for example, is a software company that works with retailers to implement automated decision-making. Blue Yonder was started in 2008 by Professor Michael Feindt, a former CERN scientist, and acquired from e-commerce giant Otto Group (an anchor investor, together with Warburg Pincus) by JDA Software Group in 2018. His company has since become the AI market leader in retail. “AI is essentially all about predictions,” he explains. “While it would be impossible to foresee tomorrow’s consumer preferences or sales levels, it is possible to make predictions about probability distributions. Once predictions have been made, they can be used to make decisions.”

Blue Yonder has worked with Otto to implement bespoke AI solutions across the brand. Hamburg based Bonprix, the group’s fashion retailer, reaches over 35 million customers in 30 countries via its digital channels and walk-in stores. Prior to their collaboration with Blue Yonder, Bonprix set prices across all its international markets using rigid price conversion tables. Now AI sets their prices automatically, optimizing for differences in markets, seasonal trends and product ranges.

Price optimization has produced greater demand and profit, says the team – particularly in Russia, where sales increased by a double-digit percentage.

AI’s decisionmaking capabilities now extend to purchasing. Each year, Blue Yonder’s AI produces more than five billion sales forecasts for Otto Group; these allow the system to manage stock replenishment by making buying decisions independently. The results? Increased sales and profit, less end-of-season waste and a reduction in product returns. As the project’s case study indicates: “With automated planning decisions, the number of returns has been minimized because orders correspond with forecast customer demand.”

Feindt points out that people often talk about AI taking human jobs away, but he does not believe this represents “employment redundancy” – instead, he says, it makes an element of human decision-making redundant.

“Technologies exist that can make human lives easier and businesses more effective; the best thing retailers can do at this point is adopt AI and give themselves 20 million fewer decisions to make every day.”

AI algorithms are also used by retailers to make decisions about individuals’ shopping experiences, explains Alex Leveringhaus, coordinator of the Surrey Special Interest Group on Ethics and AI at the University of Surrey in the U.K. They can generate complex customer profiles based on searches and buying histories.

“Based on this data, the algorithm ‘decides’ which products to show a prospective customer,” he says. “For the customer, the benefit is that they see only the most relevant products, while the retailer is able to target its customers more effectively.”

But AI’s impact doesn’t stop at sales and customer experience. Supply chains are also being reinvented. A joint DHL/IBM white paper, “Artificial Intelligence in Logistics,” published in 2018, describes the benefits and challenges of logistics becoming an “AI-driven industry.”

The report identifies myriad AI opportunities throughout logistics – from predictive demand and capacity planning to AI-powered visual inspections and the use of smart assets, such as AI-enabled robots and autonomous vehicles.

In the warehouse, computer vision and deep learning can already offer real-time inventory and shelf management solutions. And in the back office, DHL’s Global Trade Barometer tool combines AI, operational logistics data and statistical modeling to give a monthly outlook on prospects for the global economy.

Best-practice case studies from other sectors will also inform how AI is implemented in logistics. Satellite imagery company DigitalGlobe supplies high-resolution images to ride-sharing company Uber. The system can decipher incredible detail, including new road surface markings and street scale changes. The data provides input for intelligent route optimization to streamline pickup, navigation and drop-off.

“This level of detail from satellite imagery can provide valuable new insight to planning and navigating routes not only for the transport of people but for shipments as well,” say the report’s authors.

Investment in AI is surging. In 2016, companies – from startups to tech giants – invested between $26 billion and $39 billion in AI, yet the report notes that adoption remains low, with 41% of firms being “unsure of the benefits of AI.”

The authors conclude that the “network-based nature” of logistics makes it a natural fit for AI. Industry should embrace AI to ensure a “predictive, automated and personalized future,” they say, adding a warning: Companies who do not adopt AI risk being left behind by competitors who do.

In the private sector, profit tends to drive the implementation of AI. In the public sector, efficient use of public funds is a key motivator. A report by the U.K.’s Institute for Public Policy Research (IPPR) states that adoption of “full automation,” including repetitive tasks and decision-making, could save the National Health Service (NHS) £12.5 billion (€14.1 billion, or $16.9 billion) per year.

Intelligent technologies are already being embraced by the NHS. At John Radcliffe Hospital, Oxford, an AI system called Ultromics is being used to improve diagnosis of cardiovascular conditions with more than 90% accuracy. And in London, leading AI company DeepMind, part of Google’s Alphabet group, has teamed up with Moorfields Eye Hospital to develop an algorithm that analyzes eye scans, then decides how patients should be referred. It has an accuracy rate of 94%, matching that of an expert clinician, says the team.

In the U.S., researchers at Houston Methodist Hospital have developed AI software that uses natural language processing (NLP) algorithms to translate mammographic findings into diagnostic information with 99% accuracy. Manual review of 50 charts took two clinicians 50-70 hours, the team asserts, while AI reviewed over 500 charts in a few hours – a saving of 500 physician hours.

AI can also streamline the management of vast quantities of medical information: “It’s humanly impossible to keep up with the daily proliferation of healthcare data,” says the team behind IBM’s Watson Health. A connected healthcare ecosystem that uses IBM’s cloud, AI and machine learning (ML) capabilities, Watson Health makes global medical knowledge accessible and shareable. Additionally, it allows this data to be analyzed computationally for use in an array of tasks from diagnosis and predicative modeling to decision-making.

The system is based on an enormous volume of data, including over 100 million patient records, 30 billion images and 40 million research documents. The data focuses on six key areas (such as imaging, oncology and genomics, and government) and users include hospitals, insurers, researchers, and patients around the world.

Mental well-being is as important as physical. Here too, AI has a vital role to play. The U.S. startup X2 AI, for example, created a chatbot called TESS. Accessed via online messaging or SMS, TESS provides immediate 24/7 access to supportive coaching for people struggling with mental health issues. Similarly, the Woebot app mixes AI-based chatbot technology and CBT (cognitive behavioral therapy) techniques. According to the design team of psychologists, data scientists and engineers, Woebot aims “to make mental health radically accessible.”

AI can also teach us a few things. U.S.-based Mandel Communications provides coaching in communication skills ranging from public speaking to work-based interaction. Mandel’s Orai app sets challenges for its users – such as the “elevator pitch” or making excuses – and evaluates their speech, giving feedback based on several indicators. “Learners have the ability to practice the delivery of any communication, and Orai’s AI instantly analyzes and scores their clarity, pacing, clutter words, energy and tone,” explains Diane Burgess Faber, Chief Learning and Design Strategist at Mandel.

In the future, Leveringhaus believes, AI will find its way into an increasing number of domains and make increasingly complex decisions. There are two potential trajectories, he says. In the first, AI will team up with humans to make decisions. “Here, the goal is to maximize the decision-making strengths of AI and the human individual, while trying to minimize their respective weaknesses,” he explains.

In the second, AI will make complex decisions without any human interference after the programming stage. “Here, the question is how to demarcate the domain in which AI is allowed to make decisions, determine acceptable levels of risk, and prevent potentialspillover effects into other domains.”

It is also notable that some of the world’s best-known tech pioneers, such as Bill Gates, the late Stephen Hawking and Elon Musk, have expressed grave concerns about AI. Together with several thousands of experts they signed an open letter on artificial intelligence in 2015, calling for concrete research on the societal impacts of AI. Musk memorably stated in an interview with CNN that: “Humanity’s position on this planet depends on its intelligence, so if our intelligence is exceeded, it’s unlikely that we will remain in charge of the planet.”

Industry and society at large now face many questions. How can we ensure that the initial programming of AI is unbiased? Will AI have too much access to knowledge? How these issues will be addressed is not yet clear. But, says Leveringhaus, “The only thing one can say with some certainty is that a dialogue between lawmakers, technologists and society at large is needed and desirable.”

AI looks set to play an increasingly vital role in how we run our businesses and live our everyday lives. Time will tell 

Published: November 2019

Images: iStock