AI can also drive down costs using sensor data to automatically restock parts instead of referring to inventory logs and by recommending predictive maintenance as opposed to reactive maintenance, periodic maintenance, or preventative maintenance, extending the life of assets and reducing maintenance and total cost of ownership. McKinsey estimated cost savings and reductions could range from 5 to 12 percent from operations optimization, 10 to 40 percent from predictive maintenance, and 20 to 50 percent from inventory optimization.
Energy
In the energy sector, downtime and outages have serious implications. One study estimated that more than 90 percent of U.S. refinery shutdowns were unplanned. A McKinsey’s survey found that, due to unplanned downtime and maintenance, rigs in the North Sea were running at 82 percent of capacity, well below the target of 95 percent, because, although they had an abundance of data from 30,000 sensors, they were using only 1 percent of it to make immediate yes-or-no decisions regarding individual rigs.
In December 2017, a hairline crack in the North Sea Forties pipeline halted production that cost Ineos an estimated £20 million per day. In contrast, Shell Oil used predictive maintenance and early detection to avoid two malfunctions, saving an estimated $2 million in maintenance costs and downtime.
Banking and investments
The finance sector is blessed, or cursed, with both a super-abundance of paperwork and a surplus of regulation. I say “blessed” because the structured nature of the data and tightly defined rules create the perfect environment for an AI intervention.
Credit worthiness: AI can process customer data, such as credit history, social media pages, and other unstructured data, and make recommendations regarding loan applications.
Fraud prevention: AI can monitor transactions to detect anomalies and flag them for review.
Risk avoidance and mitigation: AI can review financial histories and the market to assess investment risks that can then be addressed and resolved.
Regulatory compliance: AI can be used to develop a framework to help ensure that regulatory requirements and rules are met and followed. Through machine learning, these systems can be programmed with regulations and rules to serve as a watchdog to help spot transactions that fail to adhere to set regulatory practices and procedures. This helps ensure real-time automated transaction monitoring to ensure proper compliance with established rules and regulations.
Intelligent recommendations: AI can mine not just a consumer’s past online activity, credit score, and demographic profile, but also behavior patterns of similar customers, retail partners’ purchase histories — even the unstructured data of a customer’s social media posts or comments they’ve made in customer support chats, to deliver highly-targeted offers.
Insurance
Some in the industry think that factors unique to insurance — size, sales channel, product mix, and geography — are the fundamental cost drivers for insurers. However, a McKinsey survey notes that these factors account for just 19 percent of the differences in unit costs among property and casualty insurers and 46 percent among life insurers. The majority of costs are dependent on common business challenges, such as complexity, operating model, IT architecture, and performance management. AI can play a significant role in mitigating these costs.
Claims processing: Using NLP and ML, AI can process claims much faster than a human and then flag anomalies for manual review.
Fraud detection: The FBI estimates the annual cost of insurance fraud at more than $40 billion per year, adding $400 to $700 per year for the average U.S. family in the form of increased premiums. Using predictive analytics, AI can quickly process reams of documents and transactions to detect the subtle telltale markers that flag potential fraud or erratic account movements that could be the early signs of dementia.
Customer experience: Insurance carriers can use AI chatbots to improve the overall customer experience. Chatbots use natural-language patterns to communicate with consumers. They can answer questions, resolve complaints, and review claims.
Retail
The global economy continues to apply pressure to margins, but AI gives retailers many ways to push back.
Reduced customer churn: MBNA America found that a 5-percent reduction in customer churn can equate to a 125-percent increase in profitability. Predictive analytics can identify customers likely to leave as well as predicting the remedial actions most likely to be effective, such as targeted marketing and personalized promotions and incentives.
Improved customer experience: A 2014 McKinsey study notes that companies that improve their customer journey can see revenues increase by as much as 15 percent and lower costs by up to 20 percent. AI provides a deeper and contextual understanding of the customer as they interact with your brand. In particular, natural-language processing and predictive analytics provide a granular understanding of your customer regarding their product preferences, communication preferences, and which marketing campaigns are likely to resonate with each customer.
Optimized and flexible pricing: Predictive analytics enable a company to implement an optimized pricing strategy, pricing products according to a range of variables, such as channel, location, or time of year. The system creates highly accurate predictive models that study competitor prices, inventory levels, historic pricing patterns, and customer demand to ensure that pricing is correct for each situation, achieving up to 30 percent improvement in operating profit and increasing return on investment (ROI) up to 800 percent.
Personalized and targeted marketing: A 2016 Salesforce report found that 63 percent of millennials and 58 percent of Generation-X customers gladly share their data in return for personalized offers and discounts. Retailers are uniquely positioned to collect a range of data on individual customers, including preferences, buying history, and shopping patterns. Predictive analytics help personalize marketing and engagement strategies. A 2017 Segment study found that 49 percent of shoppers made impulse buys after receiving a personalized recommendation and 44 percent become repeat buyers after personalized experiences.
Improved inventory management: The days of overstocking inventory are quickly diminishing as retailers realize that optimized stock equals more profit. Predictive analytics gives retailers a better understanding of customer behavior to highlight areas of high demand, quickly identify sales trends, and optimize delivery so the right inventory goes to the right location. The results are streamlined supply chains, reduced storage costs, and expanded margins.
Legal
AI is tackling the mountain of paper that characterizes most legal proceedings by providing better and smarter insights from organizational data to detect compliance risks, predict case outcomes, analyze sentiment, identify useful documents, and gather business intelligence to make better-informed decisions. Through automation and the use of predictive analytics, these technologies have significantly helped reduce the time and costs associated with discovery.
A 2018 test pitted 20 lawyers with decades of experience against an AI agent three years into development and trained on tens of thousands of contracts. The task? Spot legal issues in five NDAs. The lawyers lost to the AI agent on time (average 92 minutes