Machine learning isn’t going to run companies, but it will profoundly alter how companies are run.
From Henry Ford to Elon Musk, we’ve strongly identified business leaders with the companies they run. More than that, though, we’ve continued to give those leaders much of the credit or blame for shifts in company behavior and market position. By essentially saying that the business and its leader are one and the same, we don’t leave much room for true data-driven strategy.
Of course, it takes skills and experience to ascend to the top of the organizational chart. Yet it’s hard to say how much of a rock star leader’s success comes from know-how and how much comes from a combination of expectations, accumulated status, and access to information and tools that aren’t readily available to subordinates.
As machine learning and artificial intelligence (AI) make their way into the organization, though, that’s going to change—not least by chipping away at the calcified cult of the CEO.
If we accept that the best use of AI is to supplement and enhance human thinking, we must start discussing its possible effects on how companies operate and, just as importantly, on how they’re run. As AI filters upwards, democratizing access to high-quality information and tools for decision-making at every level of the business, top executives will be under increasing pressure to use AI solutions to deliver extraordinary value.
AI will not be the end of leadership, but it will change what’s needed to be a successful leader. CEOs will leverage the ability of AI to turn massive amounts of data into answers to complex strategic questions at a massively accelerated rate.
In fact, AI will let them ask questions that weren’t previously economical to answer, as well as questions they didn’t even know to ask. As other top executives also turn to AI to inform their input into corporate strategy, the effect will be amplified across the entire C-suite (collectively known as CXOs).
Once AI makes leading from the gut look careless instead of commendable, CXOs won’t be able to justify idiosyncratic management styles and questionable decisions by chalking them up to innate, instinctive leadership skills or past successes. They’ll still need charisma, but they’ll need to temper it with the ability to connect through vision and humility that makes them someone to respect rather than fear. And in addition to emotional intelligence (for more, see “Leadership: It’s All in Your Mindfulness,”) they’ll need to combine strong strategic thinking skills with increasingly sophisticated analytic tools to help them rally the organization.
AI’s presence in the C-suite will mean less reason to admire leaders for having the right hunches and more reason to applaud them for asking the right questions. But first they’ll need to learn what the right questions are.
C-Suite executives in the AI era will become masters of inquiry.
Lone wolves, C-suite execs who use instinctive leadership skills or past successes to make decisions, must become evidence enthusiasts, who combine AI tools with emotional intelligence and strategic thinking to emphasize inquiry over gut thinking.
|Lone wolves||Evidence enthusiasts|
|Past experience||Perpetual learning|
|Gut instinct||Data-driven decision-making|
|Reputation for strong leadership||Emotional intelligence and ability to inspire|
|Expectations and preconceptions||Patterns in data|
|Short-term strategy||Long-term vision and purpose|
Algorithms aren’t likely to replace humans at the top of the organizational chart in the foreseeable future because AI is a long way from even approximating the human ability to solve problems that aren’t well defined. You can’t simply ask the AI, “Who are my top-performing regional managers?” You must teach the algorithm all the criteria you use to define performance and your thresholds for high performance.
Once it knows what it’s looking for, though, AI is excellent at identifying patterns in masses of data and using those patterns to build the kinds of complex insights humans can use to inform their decisions. That’s where the disruptive promise of AI in the C-suite will lie.
Of course, top executives sometimes grapple with ultra-high–value choices like “Should this media empire merge with that one?” Most companies wouldn’t and shouldn’t be comfortable applying machine learning to these one-offs. They simply don’t have enough relevant data for the AI to generate meaningful results.
But most C-suite decisions aren’t one-offs; they recur over time. And as they do, the AI will compile a vast trove of past data that will inform decisions about critical issues, like competitive intelligence, finance, compliance, staffing, physical operations, and supply chain optimization.
In the AI-enabled C-suite, hunches and gut instincts will still be important, but they will be tools for inquiry rather than action. AI will enable CXOs to prove or dispel their hunches with extensive data and analysis rather than relying solely on personal experience or counsel from trusted advisers.
Using natural language processing, they’ll be able to type or even say something like “Show me last year’s revenues by quarter and region, and break it out by product categories” and get results instantaneously.
This is likely to happen in the near future, as implementation will essentially be a matter of adding machine learning functionality to existing data warehousing technology.
AI will also eliminate a major frustration even among those CXOs who aren’t prone to managing by gut: the need for timely and relevant data analysis. Many organizations struggle with long lead times for analyzing data as demand for fast decisions increases.
For example, today’s CMO needs to wait weeks or months for the marketing department to field and analyze a survey of customers before accurately gauging the success of a new product. With AI constantly monitoring inputs such as purchase data, search traffic, and social media, the CMO will be able to track and respond to customer sentiment in real time.
Similarly, a board of directors could use AI to compile all the relevant information about a potential acquisition and keep it up to date in real time, accelerating decision-making about the acquisition by ensuring that every board member has the most relevant information at a moment’s notice.
One of the first areas where AI is likely to emerge as an indispensable companion to top-level managers is compliance. “The C-level executives are the ones who evaluate the costs, risks, and implications of noncompliance,” says Surendra Reddy, founder and CEO of Quantiply Corporation, a company that creates AI-powered compliance applications.
“AI gives them key performance indicators that help them decide the key factors,” Reddy says. “Then AI puts those metrics into different contexts so top management can see what is happening, what might happen, what will happen, and what has happened. The point is continual optimization so you can act on those intelligence vectors.”
Another C-suite issue for which AI will be invaluable is human resources. Staffing, with its thousands of people from whom to gather data, lends itself to the clear decisions AI excels in, such as “Will this person’s specific skills be a good fit in this specific job opening?” or “What additional training does this employee need to be ready for a promotion?”
In the C-suite, AI will give the chief HR officer greater visibility into the workforce for better control over employee training, engagement, diversity, and succession planning.
It will also help leaders assess available talent to determine how best to set up a team, break into a new market, or staff a new location, such as by comparing the potential costs and benefits of hiring from outside to investing in training and redeploying existing employees.
Moreover, AI could prove invaluable for improving retention and preventing turnover by allowing the company to use past data about departed employees to predict not just who’s most likely to leave but also why. It can then change processes and management at even the highest level to disrupt those patterns.
One of leadership’s main roles is to lead change, but change is hard and change management is rarely done well. In the era of machine learning, it’s not hard to imagine that executive leaders could use this technology both to ensure that a change is worthwhile and to help nudge employees to be more receptive to it.
“All-hands meetings and company memos are going to look like Stone Age tools in the age of AI,” says Roy Bahat, head of Bloomberg Beta, the early stage venture capital fund backed by Bloomberg LP to focus on AI and machine learning in the workplace.
But do companies have enough of the right kind of data to use AI to manage internal change? And even if they do, is it a good idea?
AI can definitely help companies personalize their approaches to inducing change, particularly when it comes to adoption and use of new technologies, including AI itself, says Param Vir Singh, Carnegie Bosch Chair and Associate Professor of Business Technologies at the Tepper School of Business, Carnegie Mellon University.
IBM recently demonstrated its Debate AI, which won a recent debate with a human expert and changed the minds of many debate observers about the topics discussed. “Companies could make use of such systems and get them to communicate with their employees with a goal of convincing them,” says Singh.
However, business must pay careful attention to transparency, confidentiality, and personal data so employees don’t start to feel like they’re being manipulated.
“It’s a change in mentality about where we are and aren’t willing to use [this technology] and the boundaries of privacy in the workplace,” Bahat says. “We have to do things like this openly and transparently and with your employees’ full knowledge. The idea that you can use AI to control your employees is nonsense.” (See “How Did AI Get There?”)
The more top executives use AI to influence or evaluate corporate performance, the more important it is for them to explain how the algorithm works, why the company is using it, and what recourse employees have if they disagree with the results.
There is no good excuse, and no good amount of time, to withhold this information. Employees, not to mention investors, analysts, and boards of directors, must be confident that AI-supported decisions are consistent, accurate, and, above all, fair.
C-suite executives are the stewards responsible for making sure the use of AI is under control throughout the organization. If a company deploys an AI-enabled digital assistant for human resources, it’s top management’s responsibility to inform employees that HR will be using the technology.
That’s because AI will need to listen and watch closely to offer insight of any real value: taking in employee–manager conversations about career development, parsing and interpreting the content, and automatically generating recommendations for career development while recording the exchange in the employee’s file.
That’s a lot of highly sensitive and confidential information.
Insisting that the entire company take an opt-in approach to such tools is a respectful policy that furthers corporate goals.
On the other hand, using the digital assistant to gather this kind of information without employees’ knowledge and explicit permission—or to scoop up information beyond the purview of HR—would be blatantly invasive and a potential reason for questioning the C-suite’s judgment, no matter how useful the patterns the AI revealed.
In short, even if AI is great at doing things humans don’t have the time, capability, or capacity to do, don’t ask a machine to do something that would make employees uncomfortable if a human did it.
The shift from imposing and enforcing change through intimidation to supporting and guiding it through solid data is fundamental to AI’s ability to make a difference in the C-suite. The higher you go on the organizational chart, the more potential each decision has to influence the productivity of the entire organization, whether that’s a few dozen people or many thousands.
In using AI to help anticipate what will happen in the future and shape the company’s actions accordingly, leaders must treat machine learning systems like they treat people: by trusting them a little bit at a time, watching the outputs, and adjusting as they go.
That loop of human wisdom is ultimately why AI and human decision-makers must work as a team.
“In the C-suite, that means adopting the stance of perpetual learning, constantly experimenting with the tools to see how they surface surprising new insights,” Bahat says. “We should think of AI as affecting the entire value chain by reducing the cost of making a prediction, but you may not know what that means for you until you actually start playing with it.”
Because so few decisions at the C-level have the clearly defined outcome criteria AI requires, making AI and machine learning useful at the highest levels of management will require companies to start defining their desired outcomes as clearly as possible.
For example, imagine that a company is launching a new product and wants the manufacturing process to be carbon-neutral. The AI will have to consider not only the relative costs of building a production line versus outsourcing manufacturing to a third party but also the relative carbon emissions of each choice, as well as whether and how those costs and emissions might change based on the location of the new facility or the identity of the outside vendor.
The company will need to identify and provide the AI with all the relevant variables, as well as guidance on how to rank those variables to determine which option is best. Otherwise, it risks results that tell the company its best choice is to do what it’s always done and get the same outcome it’s always had.
This may be the biggest risk in using AI to support decision-making, especially at the C-level. If an organization is less than crystal clear about its desired outcome, from finding the best location for a new plant to creating greater gender diversity in the executive pipeline, AI’s results could be useless—or worse, actively damaging.
AI and machine learning models can be hard to test to ensure that they work as intended, which makes it easy to accidentally create an algorithm that reinforces or even creates bias, warns Singh.
“If you use an algorithm that leads to even an unintentional bias, that could have negative implications for the organization,” he says. “This is particularly important when using AI at the C-suite level.”
It’s not always possible to know whether a question that can be addressed by AI is worth asking. As AI becomes more available, affordable, and sophisticated, though, inquiry will become economical in many more cases.
Indeed, CEOs will be able to ask more questions that were once too complex to answer and to resolve questions that might not previously have been answerable in an analog world.
Surendra Reddy, founder and CEO, Quantiply Corporation
What’s more, answering those questions with AI and machine learning will have an enormous ROI, given how fast the cost of the technology is dropping.
Optimizing business processes by automating them with AI will free the C-suite to focus time and energy on more strategic decisions. Here, too, AI will be useful in what Reddy calls “the corporate equivalent of war games.”
In other words, he says, AI that uses past data to make recommendations about possible alternatives will let top managers test many different scenarios and determine how best to adapt business processes to manage risk across functions for any or all of those potential outcomes.
“AI will change the C-suite into the control room, the situation suite guiding the AI to game out scenarios and make decisions,” Reddy adds.
“The enterprise will become smarter, nimbler, more responsive, and more engaged with customers as it captures more information and incorporates it into the decision-making process,” he says.
The most consequential question investors can ask is “Where should we put our money next?”
Bloomberg Beta didn’t want to wait for entrepreneurs to make the first move. Instead, it launched its Future Founders Project, which applied machine learning techniques to data about company founders to find people in the technology community who had yet to start a company but were likely to do so if encouraged.
The firm worried that using existing data about company founders would replicate existing patterns, but as Bloomberg Beta head Roy Bahat explains, the process actually unearthed many more promising candidates than expected.
What’s more, many of those potential founders were among people over 40 and women, two groups that conventional wisdom says are less likely to start companies.
“We’re investing in AI because it can perform these tasks that are both beyond and beneath executives, tasks for which humans would never have the time or ability,” Bahat says. “Like finding correlations between different economic variables or using satellite imagery to understand the current state of a market. Or proactively identifying people who might be good at starting a company.”
We can’t say yet what the ROI of AI at the executive level will be. Until it’s implemented, we won’t know what new patterns it will surface in existing data or how those patterns might lead to improved productivity, greater efficiency, or cost savings.
However, AI will likely influence almost any decision a CXO can make. It won’t just deliver more data and informed predictions about how new initiatives might influence the organization. It will let CXOs see how those initiatives might be tweaked to have a more positive impact.
Far from simply being another layer of technology, AI will usher in a new era of leadership. Leaders will need analytic skills rather than accumulated knowledge; they’ll need an ability to inspire rather than control; and they will use AI-driven input to create a long-term vision and purpose for the organization rather than a short-term strategy.
Instead of a cult of personality at the C-level, we’ll start to see a cult of competency—one where what matters most is not the individual in charge but what the entire C-suite can do with all the information at its disposal. D!
About the Authors
Brigette McInnis-Day is the EVP of Human Resources at SAP. She leads the People and Organizational Strategy for the Global Customer Operations Business, which is SAP’s customer-facing organization.
Steven Hunt is the Senior Vice President of Customer Value at SAP. He is responsible for guiding the strategy and deployment of knowledge, tools, and process improvements that increase the value customers receive from SuccessFactors and SAP Cloud software as a service solutions.
Madhur Mayank Sharma is Director and Head of Machine Learning for HR products at SAP.
Dr. Markus Noga is vice president of Machine Learning at SAP. Machine Learning (ML) applies deep learning and advanced data science to solve business challenges. The ML team aspires to building SAP’s next growth business in intelligent solutions and works closely with existing product units and platform teams to deliver business value to their customers. Part of the SAP Innovation Center Network (ICN), the Machine Learning team operates as a lean startup within SAP with sites in Germany, Israel, Singapore, and the United States.
Ever since discovering the fledgling internet in the early 1990s, Fawn Fitter has been fascinated by the places where business and technology intersect. She’s spent 15 years in San Francisco, watching the ebbs and flows of the digital economy and writing for magazines, including Entrepreneur and Fortune Small Business.
Read more thought-provoking articles in the latest issue of Digitalist Magazine, Executive Quarterly.