A great article from Harvard Business
Review on how to integrate AI in a company. This should be compulsory reading at many board meetings. AI is a moving target so understanding what exactly it can do for your organization may be illusive. But understanding how to integrate it within a company is not only possible but a necessity. Within a short few years the gap between companies who have done it successfully and the rest will be unbridgeable.
Too many business leaders still believe that AI is just another ‘plug and play’ incremental technological investment. In reality, gaining a competitive advantage through AI requires organizational transformation of the kind exemplified by companies leading in this era: Google, Haier, Apple, Zappos, and Siemens. These companies don’t just have better technology — they have transformed the way they do business so that human resources can be augmented with machine powers.
How do they do it? To find out, we conducted a multistage study over five years, beginning with a survey of senior managers and executives, followed by interviews and surveys across a wide range of industries to identify technology implementation strategies and barriers, and in-depth studies of five leading organizations. Our key takeaway is counterintuitive. Competing in the age of AI is not about being technology-driven per se — it’s a question of new organizational structures that use technology to bring out the best in people. The secret to making this work, we learned, is the business model itself, where machines and humans are integrated to complement each other. Machines do repetitive and automated tasks and will always be more precise and faster. However, those uniquely human skills of creativity, care, intuition, adaptability, and innovation are increasingly imperative to success. These human skills cannot be “botsourced,” a term we use to characterize when a business process traditionally carried out by humans is delegated to an automated process like a robot or an algorithm.
How do leaders get the most out of AI?
From our research we have developed a four-layer framework that shows organizational leaders how they can create a human-centric organization with super-human intelligence. The four layers are not “steps,” which would imply a sequential progression. The four layers of intentionality, integration, implementation, and indication (the Four I model) must be stacked all together, or else the use of AI will fail to deliver a sustainable competitive advantage. Here’s how it works.
The first layer of the Four I model is intentionality of purpose, beyond the mere pursuit of profits. An intentional organization knows why it matters to the world, not just its shareholders. A good example of intentionality in the use of AI comes from Siemens, which evolved from a shareholder-profit-maximizing power generation and transmission company into a leading provider of electrification, automation, and digitalization solutions with energy-efficient, resource-saving technologies driven by AI and the Internet of Things (IoT) in service to society. This cultural shift toward a higher human-centric purpose impacted not just marketing and product design but also the strategic decision to, as Scott D. Anthony, Alasdair Trotter, and Evan I. Schwartz wrote for HBR, “divest its core oil and gas business and redeploy the capital to its Digital Industries unit and Smart Infrastructure business focused on energy efficiency, renewable power storage, distributed power, and electric vehicle mobility.” While financial performance and shareholder value will always be important, creating human-centered, technology-powered organizations will actually drive financial performance in the age of AI.
To that end, Siemens is launching a combination of hardware and software that enables AI throughout its Totally Integrated Automation (TIA) architecture, an approach that aligns Siemens’ mission with its AI strategy. The TIA architecture uses AI as a bridge that spans from corporate headquarters out to industrial end users. Siemens’ proprietary “MindSphere” is a cloud-based IoT operating platform that reaches into Siemens’ industrial user-operated controller and field device products. The MindSphere’s neural processing unit module allows human users to benefit from Siemens’ in-house AI capabilities, while also enabling human users to impart their own experience to train the machines. According to Siemens Factory Automation specialist Colm Gavin, “With artificial intelligence we are able to train, recognize, and adjust to allow more flexible machinery. Because, do we want 10 machines to package 10 different types of products, or a tool that accommodates different packages and different sizes and automatically adjusts to the new format?” Smarter machinery with TIA architecture leverages AI to advance the company’s intentionality, while increasing flexibility, quality, efficiency, and cost-effectiveness for its end users.
Alternatively, a negative example of the relationship between intentionality and AI is illustrated by recent issues confronting Facebook. Facebook’s mission, “to give people the power to build community and bring the world closer together,” sounds noble. Yet recent use of its AI has raised concerns from advertisers and civil rights groups alike. The social media giant has struggled to align its mission with its use of AI that seems to have the opposite effect: Facebook’s content “feed” is driven by algorithms that prioritize inflammatory, misleading, and socially divisive content. Facebook’s use of AI seems to drive social division, which is antithetical to its purpose as a social media company, and is having financial consequences. Because its algorithms have promoted disinformation, violence, and incendiary content, major advertisers are now cutting ties with Facebook, dealing a strong blow to the company that derives 98% of its income from ad revenue. Some of the largest brands in the world, including Coca-Cola and Unilever, pulled advertisements from Facebook for promoting content antithetical to their brand’s values, resulting in a one-day drop of 8.3% in market value, or $56 billion.
The second layer of the Four I model is integration of human and AI resources across the organization. To lead in the technology era, companies must shift away from silos to organizational structures with flexible teams that integrate people horizontally and vertically, from product creation to strategic decision making. As one executive we spoke with explained, before the AI shift, it was necessary for workers to have deep knowledge of a narrow area. Today, deep analytical content can come from AI. What is needed is the ability of workers to synthesize information, which means collaborating across functions and working in cross-functional teams. To foster innovation and adaptability, organizations need to transition from rigid hierarchies to flexible, agile, and flatter structures. Google, Haier, and Zappos may have differences in their organizational structures, but the common elements are flatness and fluidity. The recommended structure is more like a playground for smart, talented people to generate customer-centric products. Employees have fluid roles in cross-functional teams around problems as opposed to individual roles and responsibilities. These teams spontaneously form when problems arise, then dissolve when the work is done, reallocating human resources as needed.
The other side of this — which can easily be forgotten — is that human and AI teams should also be structured in an integrated manner. This allows humans to transcend their ordinary cognitive limitations, without placing unreasonable reliance on a robot to perform human tasks that require high degrees of care and skill. An example comes from the medical context, where AI offers tremendous potential not as a substitute for, but as a supplement to, physician-driven care. Recent research in the journal Nature found that, “good quality AI-based support of clinical decision-making improves diagnostic accuracy over that of either AI or physicians alone.” This means high-stakes, highly-skilled human decision-making can benefit from AI so long as it is integrated properly within the human decision-making context.
The third layer of the Four I model is implementation. Implementation requires engaging human talent, tolerating risk, and incentivizing cross-functional coordination. An executive at a large pharmaceutical we spoke with said, “you have to get people to believe in the technology.” We saw this in another of the companies we spoke with when we learned that despite having integrated AI, managers were modifying the output values from the algorithm to fit their own expectations. Others in the same company would simply follow the old decision-making routine, altogether ignoring the data provided by algorithms. Therefore, human behavior is central to implementing AI.
Top performing companies spent significant time communicating with employees and educating them, so that the human talent understood how machines made their jobs easier, not obsolete. To build trust in AI, it is imperative for leaders to communicate their vision transparently, explaining the goal, the changes needed, how it will be rolled out, and over what timeline. Beyond communication, leaders can inoculate their workforce against fear of AI by arranging for visits to other companies that have undergone similar transformations, providing a model for workers to see with their own eyes how the technology is used.
We saw many approaches to this in our research. Pilot projects where technology is rolled out in a limited scope give workers some ownership over the adoption process. Giving workers an opportunity to tinker with the technology before a final adoption decision is made eases the transition. Financial services firm Capital One even created an internal training institute called Capital One University that offers professional training programs to promote a broader understanding of analytics throughout the organization’s culture.
The fourth layer of the model is indication or performance measurement. Ultimately, success and progress need to be measured, and leading companies have moved from traditional productivity measures to aspirational metrics. Using the right indicators can drive improvements and help a business focus on what they deem important. Aspirational metrics that incentivize innovation and creativity encourage employees to exercise those uniquely human traits. The lesson is to be careful what you measure. Monitoring the wrong performance indicator has a strong tendency to lead to the proverbial tail wagging the dog. Humans are clever, and if incentives are not properly aligned with intelligently designed performance metrics, human workers will resort to lazy, clever, and cynical hacks to game the system, maximizing the appearance of performance under one measure while actually failing to deliver the output that management was actually hoping for when they implemented that measure.
Most companies use KPIs, but in our research we saw that successful companies more often used Objectives and Key Indicators (OKRs). What we learned was that KPIs by themselves don’t encompass strategic and ambitious goals needed in the age of AI and they don’t motivate to reach for the sky. The goal of OKRs is to precisely define how to achieve ambitious objectives where failure is imminently possible, through concrete, measurable specifications. They encourage creative, novel, and aspirational performance by showing progress toward a goal even if the goal itself is unattainable. Google famously started using OKRs in 1999; a change some even credit as a critical element of Google’s success. At Google, OKRs have helped develop transparency. Everybody knows the company’s goals, what everyone is doing, how they have done in the past, the trajectory they are on, and how they are getting to where they want to go.
Building Companies on Super-human Intelligence
Our research shows that AI is so much more than just the latest incremental improvement in existing technology, however deploying it effectively takes leadership and coordination across all sectors of a company. Unlocking the full potential of an organization’s human resources by adopting AI strategically requires revisiting the very structure of the company and how it measures its progress toward fulfilling its mission. These issues are core issues to the identity of a company and modifications here are fraught with insecurity and risk, but this is a risk needed to compete in the age of AI. Intentionality, integration, implementation, and indication must be layered in order to create a human-centric enterprise governed by super-human intelligence. Achieving this requires talent at all levels to have systems-thinking, understand how the work being done meshes with that of others elsewhere in the organization, how it meets customer needs, and how it impacts the company’s strategy and financial picture. By following the Four I model, companies can unlock super-human intelligence without losing the human touch.
We were surprised to discover how few organizations have unlocked this secret. But we were encouraged by the progress of the ones that had. With this model, we hope, more companies can create the conditions for realizing super-human intelligence and performance, delivering sustainable competitive advantages in the age of AI.
No comments:
Post a Comment