AUGUST 15th 2017
Top 7 Insights for Manufacturing Execs from McKinsey's 2017 AI Report
By: Annie Wu
The McKinsey Global Institute (MGI), named the 2016 number one private-sector think tank in the world, gives executives data and insights to make better decisions. Their reports have covered more than 20 countries and 30 industries, and they recently publish the 2017 annual report on Artificial Intelligence (AI).

The research is exciting because it shows firms that proactively adopt AI have a significantly higher profit margin than partial AI adopters or non adopters. It's also 80 pages long, so we distilled it down to 7 insights that can help manufacturing executives, and some practical advice on how firms successfully implement AI.

1. AI will improve R&D outcomes

By analyzing more market data and internal data and using more complex algorithms, AI will assess whether a prototype will succeed or fail. McKinsey found that AI for R&D can increase productivity gains of 10-15% and reduce a product's time to market by 10% or more (25).

The AI startup Motiv uses AI to compress a manufacturing company's design process from years to just 4 weeks (25). Similarly, Intel achieves a 10% higher yield for their integrated circuit products than their competitor (54).

2. AI will reduce inventory requirements

Forecasting drives S&OP performance by influencing production, inventory, pricing, and replenishment decisions. McKinsey found that AI reduces forecasting errors by 30 to 50% (22).
Machine Learning forecasting systems will analyze more data than legacy statistical forecasting systems. Using historical sales data combined with real-time external data such as weather, local events, pricing, and promotion campaigns, machine learning will more accurately anticipate demand and therefore sales. Predicting demand earlier lets companies reduce inventory requirements according to demand, optimizing sales and minimizing waste.

AI will also give companies more transparency into the supply chain of their suppliers. Monitoring suppliers' availability and performance information will help them optimize their own production and inventory in real time.

Manufacturers supply thousands of SKU's to thousands of retailers. AI can keep manufacturers stock and balance inventories in real time without sacrificing supply chain transparency. AI can reduce missed orders by up to 65% and reduce overall inventory by up to 50% (22).

3. AI will improve pricing decisions

Manufacturers will also benefit from AI's ability to optimize pricing. With accurate forecasts and optimized inventory, AI will price goods dynamically, lowering prices when there is less demand or price inelasticity. Manufacturers will use AI to reach revenue and goals more consistently with dynamic pricing and promotions.

4. AI will reduce production errors

Many manufacturers already use autonomous robotics and technology. Swisslog reduced stocking time by 30% since their implementation of autonomous vehicles in their warehouse (43). Siemens' Electronic Works Amberg is a virtual factory where nearly 75% of their entire production process is automated with results of their logic circuits being 99.99988% defect free (53).

AI will reduce assembly inefficiencies within the production process, which cost manufacturers billions of dollars annually. AI will outperform current fault detection systems by using real time information to continuously update the process predictions giving manufacturers visibility into component availability to continually manage risk. Moreover, AI will automatically predict, assess, and respond to disruptions and unplanned errors and optimize scheduling.

5. AI will reduce delivery and after-sales costs

AI advances lead to more automated experiences. AI already optimizes deliveries. A semiconductor manufacturer reduced their delivery time by 30% and improved production yield by 3 to 5% by using AI to determine the best time to make deliveries (55). AI can reduce warehousing and transportation costs by 5-10% (22).

AI will also help companies optimize after-sales service, an important component of the manufacturing value chain. GE Aviation offers their planes on an hourly model, where flight operators only pay for hours of flight. AI will help manufacturing companies provide maintenance and repairs when necessary based on complex analyses of sensor data. AI will also optimize response to unplanned maintenance events. One firm increased profit by over $300 million by using AI to forecast repairs 10 years in advance (28).

How to implement AI in the enterprise

To implement AI, smart firms identify and build a business case around a pain point in their business. It is best to focus on a project that is limited in scope, where there AI has already been proven to solve a specific problem at scale. When companies benefit from AI, they should apply it more broadly with long-term projects and experiment with unproven use cases.

Because there's a shortage of data scientists, it's challenging to hire data science talent and 'translators' who can lead AI business projects. To implement AI, some firms choose to partner with third parties like AI startups or AI consulting firms. By doing so, they reduce their costs, reduce their risk, and can run quick experiments to evaluate their ROI potential of applying AI to different parts of their business.

Firms must start small and slowly build internal capabilities, whether this means partnering with academia or other organizations, hiring outside talent, or reskilling their current team. McKinsey's Artificial Intelligence Discussion Paper acts as a guide and outlines the current state of artificial intelligence in the business world, the benefits of AI in various industries, and how firms will be able to implement their own AI solutions. Check out McKinsey's full report.

Made on