Building a Business Case for Bots? Throw Out Cost-Benefit Analysis


Wait a minute! How can a business case even exist without cost?  Isn't a business case all about financial costs vs. benefits?

If we're talking about your father's business case, the answer is yes. But take it from someone who may be old enough to be your father, times have changed, and so has the business case.

Of course a cost-benefit business case can be constructed for your AI initiative, but, frankly, only to the end of satisfying those who don't know another way. The conventional business case model is an industrial-age construct using industrial-age assumptions.

Consider the following:

  • The payback horizon for a significant AI initiative is likely to be beyond the point in time where predictions can reasonably be made, simply due to the pace of disruptive change.
  • The case for change, being based on a predicted future state versus current state, is based on a false premise that the current state otherwise would remain unchanged over the same period of time. It won't.
  • Finally, hard-dollar benefits based on FTE reductions assume that bots are replacing human beings, when the real benefits of bots will stem from their ability to do things that human beings cannot.

So, go ahead and build that cost-based business case if you must. But if you really want to serve the business I'd like to suggest a different approach.

The Value-Driven Business Case

Value is not the same as cost, nor is cost necessary to articulate value. When it comes to delivering HR services, I define value as the fulfillment of customer-valued outcomes. By "customer-valued" I mean they are states of existence for which the customer, which generally speaking is the enterprise, would be willing to pay money. An employee in possession of the correct amount of pay on pay day is an example of a customer-valued outcome, because it is a state for which the enterprise is willing to pay money. As a rule, customer-valued outcomes are expressed without verbs, forcing the thought process away from valuing activities and instead valuing desired states. I sometimes call this the genie test, because if the organization had a genie it would be ridiculous to ask the genie to perform tasks when it could simply make things be. No activities. No verbs. Just outcomes. Poof.

The value-driven business case is an articulation of how the proposed solution, in this case AI, will produce valued outcomes better than the current solution. Perhaps it is an outcome that the current solution cannot produce at all. The business decision is based on how much the organization is willing to pay for the outcome(s) in question.

Identifying Valued Outcomes

What are the customer-valued outcomes on which you will base the value-driven business case? Here is where creativity becomes critically important. If the valued outcomes you contrive are merely the outcomes currently produced by human beings, only at a lower cost, your business case will fall flat. But now that you've thrown cost out of your equation, you're free to think beyond cheaper and instead think better. What can AI do that human beings cannot do? What valued outcomes do those things create, increase or improve? And, finally, how much is the organization willing to pay for them.

The key to creating a compelling value-driven business case is creating a vision in which AI is delivering customer-valued outcomes that human beings are unable or ill suited to deliver. Is it answering questions at 3 a.m. anywhere in the world? Is it processing massive amounts of data that would take human beings days or weeks to analyze? Or is it performing mind-numbing tasks that send human beings looking for another place to work?  

Now that cost is no longer driving your business case, let your imagination run wild. What you come up with may be more compelling than any mere cost-benefit business case could ever be.

When Will AI REALLY Take Off?


There's a saying about English ivy -- first year sleeps, second year creeps, third year leaps. I can testify to its accuracy, but it's incomplete. Here in Atlanta, where the weather alternates between desert and rain forest conditions, there should be a fourth part that goes: and then completely takes over. I've seen unchecked English ivy take down giant oak trees. I think artificial intelligence is like English ivy.

In HR, we're straddling the sleeping and creeping stages of AI. In a recently completed survey of HR delivery practices, which was a research partnership between HRSSI and Aon's HR Effectiveness practice, this was made evident:

  • Whereas only about 15 percent of HR organizations currently utilize AI assistants, otherwise known as chatbots, the exact converse of 85% expect to begin utilizing them in the next couple of years.
  • Even more revealing, and the justification for my approaching claim, is that the most pressing concerns about further adoption of bots are: a) the cost of initial configuration and ongoing maintenance; and b) the cost to implement an employee experience platform that enables chatbot functionality. Much lower on the list, at the very bottom in fact, was the concern that bots could function as promised.
  • Confirming this confidence in bot capabilities, the data also showed that the ability of bots to deliver information faster and more accurately than human beings ranked higher on expected benefits than cost savings through FTE reductions.  

In other words, the chief barrier to increased AI adoption is making the business case. To illustrate, I'll refer back to the early days of human flight. We're all familiar with the old footage of early failed attempts at human flight, with brave, if slightly crazed, "pilots" launching themselves from cliffs in contraptions built to imitate the flight behavior of birds, inevitably crashing to the ground. But eventually, after enough bruises and broken bones, these painful experiments produced the first true flying machine. The rest is history. 

The mistake made by those earliest bruised adventurers was trying to fly like a bird. Human flying machines don't fly like birds, in that the don't flap their wings. They do, however, employ the same aerodynamic principles as birds' wings. When experimenters realized that humans could fly by using the same aerodynamic principles as birds but not the same behaviors, human flight, well, took off.

Airplanes still can't do many of the things birds can, but birds can't do many of the things aircraft can either. Once we realized that flight was not about flying like a bird but as a bird, we were able to create machines that in certain ways could fly much better than birds. The birds human built can carry hundreds of people across vast oceans, protect our shores from invasion and fly into outer space, things no bird can do.

Now I arrive at my main point.

In these early days of AI, we're still making the mistake the early flight explorers did in trying to make AI do what human beings do. Instead, like the later successful inventors, we need to make AI do what human beings cannot. 

As long as we think of AI as a machine that can do what humans do, business cases for AI will fly about as well as the first flying machines. It's not that they can't do some things that humans can; that's been proven already. But who really cares? Do we truly care that chatbots can answer routine questions and retrieve standard information like human agents? When you consider the volume of such inquiries and the amount of time (FTEs) spent resolving them, do you really think Wall Street analysts will pay attention? I do believe that chatbots can pay for themselves, and then some, by doing this type of work. But that's not what will make AI take flight. Instead, we must focus on bots doing what human agents cannot, which means providing services that currently are not possible through human channels.

This raises the central question: if bots aren't doing what human agents do, and thus enabling headcount reductions, how can we measure their ROI? 

Back to English ivy...

The first generation of bots, which do what human agents do, will pay for themselves, but barely more; in other words, they will creep. But the next generation of bots, built to do what human agents cannot, will deliver far more. What that might be, I frankly can't say (I don't think any of us really can). But trust me, they will. That's just how technology works.

In the interim, I see organizations deploying bots to do what human agents otherwise would do, basically breaking even or a little better.  Over time, however, bots will be utilized in ways we cannot today imagine, doing not what human agents can do but what they cannot. Consequently, business cases will no longer be based on headcount savings but things far more powerful. I can  speculate about what those more powerful things might be...more engaged employees, better-informed talent decisions, etc. But that's not the point. Don't ask your consultant what these future bot capabilities are, because we don't know. Nobody does.

Despite those first attempts at human flight being as comical as painful, they eventually took us to the moon and back again, because we didn't give up. Like English ivy, human flight slept, then crept, the leaped...and then completely took over. Bots will do the same.

Be patient, have faith, and, most importantly, take part in the experiment.


HR, Stop Saying "That Will Never Work Here"

OK, maybe it's true that such-and-such practice will not work at your organization. But I'm here to tell you that as an HR professional phrases like "that will never work here" shouldn't pass your lips. The simple reason is that such phrases presume the organization won't or can't change, and HR, if it is to survive as a profession, needs to be focused on helping the organization change, not hindering it.


Of course, HR people have been conditioned to think this way. When an employee or manager can't figure out how to complete an HR-related procedure, they go to HR. When a manager can't manage effectively or, worse, commits a policy or compliance offense, HR is called in. Through these and other situations, the organization that HR professionals most often see is one that requires babysitting or policing. 

I intentionally used the term "never" to make a point. Clearly, when challenged any reasonable person would admit to the absurdity of using the term "never.".Yet when I hear varieties of this statement uttered the word "never" is frequently used. Thus, the choice of "never" can only imply a sense of resignation; an underlying belief that the organization cannot or will not change. Again, this ought not be HR's modus operandi.

For most of its history, HR has found success in being the keeper of the status quo, indeed, I should say, the keeper of the bureaucratic status quo. Put another way, HR is the keeper of the organization's structure, a word which itself implies rigidity. This warrants a little context.

In organizational theory terms, structure is said to have three dimensions:

  • Complexity - the extent of differentiation in the organization, including specialization, division of labor, hierarchical layers, etc. 
  • Centralization - where the locus of decision-making authority lies
  • Formalization - the degree to which the organization relies on rules and defined procedures to govern behavior

Without delving into theory, it should be obvious that HR is smack in the middle of managing the organization's structure, using such tools as job descriptions, competency models, job families, compensation structures, career models and, of course, policy handbooks.

Doing without these tools entirely seems as absurd as using the term "never," if for no other reason than HR's mandate to manage legal and regulatory risk in employment matters. But, one by one, at increasing pace, organizations are abandoning conventional structures in favor of more "agile," network-centric models. And this trend doesn't just apply to unconventional companies like Zappos and W.L. Gore. A recent HBR article titled Agile at Scale describes expressions of agile organizational design at traditional manufacturing companies, like Bosch and 3M.

When organizations transform along these lines HR not only must find new ways of working, it must go further to help lead the transformation. Obviously, phrases like "that will never work here" have no place in such situations. This online article published in Business Insider tells of one such journey by Zappos' Head of People Operations, Hollie Delaney.

Slowly but surely, as organizations strive to keep pace with change in the digital era, the bureaucratic, hierarchical organizational structure is becoming extinct. Don't let it happen to you.

Natural Intelligence: a blog about HR in the digital era

Occasionally, I'm sufficiently unsure whether the online agent I'm "chatting" with is human or AI assistant (bot), I've actually asked. So far, all I've questioned have either been human or very clever human-impersonating bots. I suppose I'll never know.


I'm not technically trained for apt, for that matter, yet for some reason I spend a lot of my time learning and talking about HR technology. Twenty-four years ago, shortly after finishing my graduate management degree, I took a freelance change management assignment to pay the bills. The company, Georgia-Pacific, would be one of the first U.S. organizations to go live with SAP HCM (then called R/3) and implement an SAP-centric HR shared services operating model. That decision defined the rest of my career; thus I spend my time today learning and talking about technology driven HR service delivery functions. 

I began this blog by calling it, simply, "blog," realizing that was like naming a dog Dog, but I honestly didn't know what the blog would be about, other than technology driven HR service delivery, naturally. But my musings, the stuff of blogs, kept returning to questions about the impact of digitization on HR Service Delivery. So, I changed the name from Blog to Natural Intelligence, a playful way of contrasting the blog's contents with bot-produced content. It's this blogger's way of checking the box confirming that I am not a robot.

Not every blog post will pertain directly to digitalization and artificial intelligence (AI), I can promise you that every post will address their implications. Frankly, how could they not? We'll talk about:

  • Where AI is going (and where it's probably not);
  • What companies are doing with digital HR and IA (and what they're not);
  • The shifting vendor landscape in HRIT
  • How digitization is changing organization structure and methods
  • ...and whatever else comes to mind.

I hope you enjoy Natural Intelligence and find it worth sharing with colleagues and co-workers. If you have ideas for things you'd like discussed, or if you have a blog you'd like us to share, let me know at

Notice This Next Time You Go Through Airport Security

If you travel by plane from a relatively busy airport (I consult and live in Atlanta, so for me this is often), I'll bet you've observed workers at the start of the security line where they check to make sure you have a boarding pass engaging in various tactics to move the line along faster. I've even seen an agent urge passengers who have already been through the ID checking stage to line up more tightly so that more passengers can move through that checkpoint.  


Being a process consultant I've tried pointing out their folly a couple of times, but I've since given up. What these agents evidently cannot see, for reasons I don't understand, is that no matter how fast they move people past the boarding pass checking or ID checking stage, the line will never move any faster than the slowest part of the process, which is the X-ray process.

This is probably so obvious to you that I'm a little embarrassed writing a blog about it. But I'm using this example to illustrate a point that many people still miss. The hurry-up-to-wait syndrome I'm describing is an example of the bigger concept of local-optima thinking, or local optimization, that happens every day in almost every organization.

Eli Goldratt, author of The Goal and Theory of Constraints (TOC), spent his career helping people understand that every process, indeed every system, has a constraint which governs the speed and/or efficiency of the process or system as a whole. If that constraint is improved, another will take its place. There's always a constraint. Thus, if one wants to improve the performance of the system as a whole it's necessary to improve or "exploit" the constraint. Makes sense, right? Then why isn't everyone practicing TOC?

The reason is that in his battle for the manager's mind TOC consistently comes up against an even more powerful concept, the most powerful management concept of the last century -- BUREAUCRACY. And, when TOC faces off against bureaucracy, the latter typically wins.

You see, constraints are easy to see in security lines and even manufacturing processes, but elsewhere they're more hidden. Likewise, local-optima thinking can be hard to spot. But any time you're measuring or optimizing a part of the process instead of the end result of the process, you're engaged in local-optima thinking. 

So, where does bureaucracy fit into the story? You see, bureaucracy divides the organization into parts, specifically silos and levels, to form the familiar pyramid structure. A bureaucratic organizational structure is, by definition, a siloed hierarchy of parts. The managers of those parts are accountable for the performance of their parts, the result of which is local-optima thinking. In a bureaucracy this is normal and expected.

In a bureaucracy, local-optima thinking occurs all the way down the chain, even within a shared services organization. If your focus, and the measures of that focus, is on the performance of parts of end-to-end processes you are thinking in local-optima terms.

The alternative is to measure performance based on outcomes, but that's a topic for another blog. For now, the next time you go through the security line at the airport, see if you can notice people optimizing parts of the process that make no difference in the performance of the process as a whole. Then ask yourself the question:  Where in the part of the organization I manage am I doing the same thing?


Seeing HR Through the Eyes of an Astronaut

The HR World is Flat

Less than a decade ago, when I first began conducting HR service delivery practice surveys, fewer than 20% of shared services operations were considered global, with 80% calling themselves single-country, or domestic. Nowadays, according to soon-to-be-released findings from the 2018 HR Delivery Practices Survey, the picture has completely flipped. Today, the reverse is true, with over three-fourths of HR shared services operations structured under some type of global umbrella.

This is not a surprise.  In those early surveys respondents also pointed toward future globalization as a top priority. Today's global landscape is made up of myriad global visions that simply became reality. Borrowing the title of best selling author Thomas Friedman's treatise on globalization, the world of HR has indeed become flat.

But being global entails far more than being structured under a global organizational umbrella. Being global is an attitude and, more importantly, an aptitude. Being global is not always good, and it's never easy. In an earlier blog about Global Business Services (June 5, 2018), I talk about the inevitable administrative burden that comes with being global and for which the advantages of globalization more than pay.

Over the years, I've consulted with many global organizations. Today nearly all of my clients are global to some degree. But most I would say have not globalized. By that I mean they are still U.S. companies with operations in other parts of the world. But they still think and behave like a U.S. company, first and foremost. 

Globalization takes a long time and a lot of work to achieve. Companies globalize much as children mature, imperceptively yet profoundly. People in truly global organizations learn to live without the comfort of a strong national center. Someone raised under a strong central authority entering a truly globalized organization must feel like floating in Vertigo by comparison. The center is both itself and nothing. It is an idea, not a place (queue the Twilight Zone music).

Global Harmonization

The pot of gold for which every global HR organization seeks is global harmonization, which being the ultimate expression of centralization seems a bit oxymoronic. More than once I've been involved in global HR projects espousing the guiding principle that processes and programs will be globally standard unless there is a government or business reason preventing it. Every organization that ventures down this path eventually runs into the wall beyond which standardization cannot go. Some hit the wall earlier than others. Some before they ever get started.

Most global companies with which I've had direct involvement didn't (or don't) quite know what to do with their global-ness. They know they need to be global, but are much less certain about how to go about it. Most organizations I've known have one foot firmly planted on U.S. soil with the other limbs in Twister-like contortions trying to be in all other parts of the world, sometimes called "rest of world." These are not truly globalized organizations, rather U.S. (or German, or...) companies with a nagging global headache.

Seeing Like Astronauts


To think globally, one must learn to see like astronauts, who upon returning to earth unanimously share the jarring impact of seeing the earth from space and realizing that the boundaries we use to dissect it are mere human creations, completely invisible from outer space.

The digital world sees the earth like astronauts. Electrons zip across time zones in real time. Data cross oceans and scale mountains almost instantly. Supercomputers learn different languages so we don't have to (be that a good or bad thing). But in the human realm we see boundaries. And, like all boundaries, mere humans made them up. And eventually they stuck.

Take, for example, the way businesses divide the world into regions, typically NA, LATAM, EMEA and APAC. Really? We put Europe, Middle East and Africa in a single region and called it EMEA? (By the way, if your company has done a bang-up job with HR shared services across all those countries please send me an email, because I'd love to blog about it. But I won't hold my breath.)

Let's put on our astronaut glasses. Let's watch the lines disappear and infinite possibilities emerge. In the digital realm, the world is just a big shapeless network lighting up in different places at different times like the firing of brain synapses watching a great movie. Why, then, shouldn't we ponder how to design the optimal network of HR professionals to support the great network of people that is the company?

The Digital Enabler

Technology has always been about breaking barriers and overcoming limitations. Whether we were to sail across the sea, fly above the clouds or transform society, technology provided us with the means.  

Technology is ever calling to us with new ideas, if we are able and willing to listen. I believe we've yet to figure out how HR services can best serve the global networks we support. I don't have it all figured out, but I have some ideas. Let's have a conversation about it, shall we?

The Essence of Design Thinking


I was struck recently, looking at college brochures for my graduating daughter, at a common design feature in many of the college campuses admired for their beauty. In so many cases, stone or concrete walking paths zig-zag across their outdoor spaces. I was thus reminded of the story I heard long ago of a new college president being given a tour of the campus grounds by the chief groundskeeper. Upon reaching the central quadrangle the groundskeeper grimaced at a foot path cutting across the manicured lawn and apologetically explained that no matter how many "keep of the grass" signs he posted, students continued to ignore the provided sidewalk. The president famously replied, "Then perhaps you should install a sidewalk where they're walking."

Perhaps this story went viral in the college groundskeeper network, but for whatever reason colleges seem to have taken the notion to heart, because it is clear that the seemingly random pathways installed on America's most beautiful campuses clearly are placed, yes, where students would want to walk. What a concept!

It occurs to me that the same shift in thinking -- outside-in thinking, to be specific -- is now happening in designing employee experiences using versions of a technique called design thinking. Despite efforts to complicate the concept so as to turn it into a consulting methodology, design thinking is essentially nothing more than putting the employee experience at the center of design, just as colleges place student behavior at the center of campus design.

Outside-in thinking is simply that -- starting with the experience, determining the best processes to deliver that experience, and then building the right structure to enable those processes. Though this seems common sense, the course we in the HR delivery model business have been following is exactly the opposite. We start with the structure we desire, design processes to fit that structure, which in turn define the experience employees will have. Design thinking stands this process on its head.

But what's too often missing in design thinking is the most important thing of all -- empirical study. In the campus groundskeeper story the evidence was a worn foot path, proof positive that students had chosen this as the desired course. Designing employee experiences takes a little more effort. In fact, without empirical study design thinking is little more than lip service using assumptions, not evidence, about the employee experience.

In the 20-plus years I've been designing and running HR service delivery models, I can count on one hand the number of initiatives involving actual employee research. Instead, believing we know what employees want and need, HR has chosen to go it alone. 

I truly understand this, because employee research is time consuming and potentially risky. It not only takes a lot of time on the part of HR but takes employees away from their jobs as well. And, if you ask employees what they think there's an inherent obligation to at least close the loop regarding how their input was used in the design. This gets rather tricky when their input is ignored. Faced with those obstacles, it's understandably appealing to make assumptions.

But don't think designing around assumptions about employee needs is bona-fide design thinking. Web site designers who have made design thinking a thing wouldn't consider going on their personal assumptions of what users and customers need and want. They do research. Like them, so should you.


Is Global Business Services Grounded or Just Taking Off?

Nothing New under the Sun

The business services version of shared services, also called multi-function shared services, has been around since the earliest days of shared services. In fact, the world's first "shared services" organization invented by GM's Alfred Sloan in the mid 1920s was essentially a global business services model. Of course, human resources didn't yet exist as a corporate function, but the function Sloan labeled General Service Staff included several functions recognizable today, including Real Estate, Industrial Relations, Traffic (logistics), Building (facilities) and others.

The business driver behind Sloan's General Service Staff function also will sound very familiar today, as indicated by these remarks from a March 6, 1924 meeting memorialized in Sloan's autobiography:

"While General Motors is definitely committed to a decentralized plan of operation, it is nevertheless obvious that from time to time general plans and policies beneficial to the Corporation and its stockholders, as well as to the individual divisions, can best be accomplished through concerted effort."

While such a statement seems quite conventional today, Sloan was espousing a major innovation that would later be taught in business curricula as the "divisional model" of organization. It can, therefore, be asserted that the business services model has been in existence as long as the divisional model has.

Why, then, in the new age of shared services, which can be traced back twenty or so years, does the business services model not dominate? On the contrary, only about one in five HR shared services models exists as part of a broader, multi-functional shared services function. And, while this percentage has eased upward slightly over the past two decades and continues to do so, the trend has been relatively flat when compared to the trend in HR shared services model adoption.

Global Business Services (GBS)

That said, there does seem to be recent uptake in the business services model. Driven primarily by globalization, there seems to be renewed interest in what is now being called Global Business Services, or GBS. 

I speculate that the GBS trend is driven less by cross-function synergies than yearning for a delivery model through which to achieve what I call "global enterprisation." Similar but different than enterprise globalization, global enterprisation aims to capture enterprise synergies on a global scale. In recent decades, it has been common for organizations to globalize rapidly, largely through acquisitions, without necessarily integrating global entities into the enterprise operating model. The result has been global diseconomy. Or, put another way, a big mess.

Reducing Global Diseconomies

For whatever reason, many of us were taught the concept of scale economies with only half the model -- the good half. In that half, unit costs decrease as unit volume increases. The curve eventually flattens out, but then, in our incomplete education, abruptly stops. But the true economies of scale curve doesn't stop. It actually turns upward, depicting an increase in unit costs with further increases in unit volume. If you don't believe it, ask yourself why the cost per paycheck at a large complex organization is greater than at a smaller, simple organization. The reason, of course, is complexity.


Looking only at half the curve, one might assume that if scale economies are a good thing then global scale economies are even better. But except in rare situations the term global economies is actually an oxymoron. Global diseconomies would be a more appropriate term. Global organizations frequently fail to achieve the benefits associated with a large organization because many are in reality a fragmented collection of smaller organizations.

Through this lens, the appeal of global business services becomes more clear. While the business services model seeks to achieve cross-functional scale economies, global business services seeks to reduce global diseconomies. It turns out, being global comes at a price, and one way seen to reduce that price is to implement a GBS model. Thus, the goal of global enterprisation is not increasing scale, rather achieving synergy. It is, simply put, about moving from being multi-national to truly global.

Go ahead HR, but Proceed With Caution

HR leaders looking to including HR processes under a GBS umbrella should proceed with an open mind as well as caution. In particular, they should be wary of pursuing the same business cases as their functional peers, like IT, finance and procurement.

HR is different than these other functions, but not merely because it deals with people and people issues, though that certainly is a factor, but because of clear differences in its operating environment. Here's what I mean:

  • Volume. In almost all cases, the customer base for HR is employees and managers. Some HR shared services organizations provide services to retirees and candidates, but these are few and far between. This, combined with the increased prevalence of self-service tools (more on this below), means inquiry and transaction volumes for HR are generally far lower than IT or finance, for instance. For example, the typical staffing ratio for tier 1 HR inquiry handling representatives is in the 1 : 1,500 - 1 : 2,000 range, meaning one tier 1 representative per 1,500 or 2,000 employees served. Very large organizations often achieve ratios of 1 : 4,000 or higher. At these ratios, it takes a very large organization to demand a large tier 1 contact center staff. If the GBS model design is geared toward achieving lower labor costs, be aware that the sheer number of HR-dedicated staff may be very low compared to other functions, making labor rate savings a relatively minor business case factor.
  • Language. For typical global organizations the most challenging aspect of globalizing HR services is language. While the spoken word is necessary across functions, processors who deal primarily with numbers and/or perform purely back-office tasks are less dependent on language skills. Many global organizations employ relatively small workforces in many countries that may have unique language requirements, often making it uneconomical to employ centralized HR service personnel who speak all the necessary languages to serve the full population. Often, language needs alone requires HR servicing staff to remain in country. Needless to say, English-only global organizations are the lucky minority in this regard.
  • Customer Base. At the risk of sounding callous, employees and managers, and even retirees and candidates, are captive audiences. Generally speaking, the organization can, with sufficient change management effort, dictate processes and procedures to them. Customers and vendors, by contrast, don't belong to the organization. The organization can't dictate which hardware and software applications they use. In these functions, the best option may be to handle inefficient procedures with cheaper labor.
  • Technology. In some ways, HR's IT landscape is well ahead of its corporate counterparts. In many organizations HR is the first, if not only, corporate function to operate using enterprise HR platforms. Another key differentiator is the fact that for most HR transactions there are self-service options in the marketplace. When employees and managers are able to process transactions for themselves, consolidating and/or offshoring processing tasks is no longer the most economical path. In fact, thanks to HR technology advances, the whole physical centralization concept is becoming a solution to yesterday's problem in the world of HR.

This is not to say HR should run away from GBS. The point, again, is that the purpose should be to enable global enterprisation, or enterprise synergy. Organizations globalize HR because they are global, not the other way around. Transaction cost savings through scale economies in HR will not excite anyone on Wall Street, so you're best to focus on a model that creates the desired employee experience and enables the company to manage talent and risk as a global enterprise.






What's the Difference Between RPA and AI?

Because artificial intelligence (AI) is relatively new to the HR space, it's understandable for there to be confusion about terms being used. In particular, I encounter the terms Robotic Process Automation (RPA) and AI assistants (i.e., chatbots) incorrectly used interchangeably. So, I thought I'd write a quick blog explaining what they are and how they're different.


To begin with, RPA and AI are indeed very different technologies, though they can be used in combination. RPA is software that can be configured to literally replicate human user behaviors, typically keyboard entries. I like to compare RPA to a player piano. A player piano uses software (previous to computerization rolls of punched paper) to literally replicate the depression of piano keys in accordance with the notes and timing prescribed by the composer. This is why it looks like the player piano is being played by an invisible person. With RPA software, the same thing occurs except instead of a piano keyboard it's a computer keyboard and the keys don't actually move. 

AI is completely different and a bit harder to explain. First, AI relies on a supercomputer, like IBM's Watson or Google's DeepMind. RPA requires no supercomputing power. The AI supercomputer runs a suite of software to perform natural language processing, data retrieval, cognitive logic and machine learning. Needless to say, AI is much more complicated than RPA.

RPA is good for highly repetitive, non-ambiguous data processing tasks, such as entering data from a spreadsheet or pre-formatted invoice into a separate software application. As long as the format of the data source, data standards and data entry procedures are consistent and routinized, an RPA application can be configured to perform the manual tasks instead of a human being. Unlike an automated upload program, however, the RPA software must replicate the actual human tasks. While they keying may be faster than a human, the same sequence of tasks must be followed. Therefore, performing the task with RPA can take considerably longer than an interface program. The advantage, however, is that RPA applications can be configured relatively quickly, almost on the fly compared to programming an interface. 

AI assistants, or chatbots, are better suited to answering questions and processing individual transaction requests. An AI assistant, for example, can retrieve a users online pay stub by asking the user for the pay period and necessary login credentials. The seemly magical power of AI is its ability to respond to requests using natural language and search available data for the most relevant response. 

I hope you found this quick explanation helpful. If you would like more detail, you might start with these Wikipedia articles. 

Wikipedia article on Chatbots

Wikipedia article on Robotic Process Automation


A Framework for Putting AI in its Place

In his 2007 Harvard Business Review article, a former IBMer named Dave Snowden published his framework for managing intellectual capital, which he called the Cynefin (pronounced kin-eff-in) Framework, borrowing the Welsh word for habitat, or place.

The framework describes five domains of knowledge:

  • Obvious
  • Complicated
  • Complex
  • Chaotic
  • Disordered (middle area)

Although his graphic representation resembles a typical four-square diagram, lacking x-y axes it is not, strictly speaking.

Moving counter-clockwise from the obvious realm:

  • The obvious domain is the realm of known knowns. It is governed by rules and the relationship between cause and effect is clear; if x, then Y. In this realm, the problem solver is required to sense-categorize-respond. That is, it must establish the facts, categorize them, then respond in accordance with established rule(s) or best practice(s). 
  • The complicated domain is the realm of known unknowns. Determining the relationship between cause and effect requires expertise, as there is a range of right answers. In this case, the bot must sense-analyze-respond. While rules still apply, the problem solver must analyze the facts to determine which rules to apply.
  • The complex domain is the realm of unknown unknowns. Rules are insufficient for understanding and the relationship between cause and effect can only be deduced in retrospect; there are no right answers. To function in this realm, the problem solver must probe-sense-respond, meaning research or experimentation is required up front to establish relevant facts.
  • In the chaotic domain events are too confusing to allow for knowledge-based response. Here, the appropriate pattern in act-sense-respond, meaning first act to establish order, then sense the impact of the action(s) and continue responding to transform the situation from chaos to complex.
  • Finally, the disordered domain is the realm of confusion. The best a problem solver can do in this domain is break the situation in to constituent parts to determine which of the other realms apply and work from there.

Clearly, bots can play a role in the obvious realm, but what is less clear is that they can play a role in the complicated realm as well. In fact, IBM Watson advertising is squarely focused on Watson's ability to solve problems in the complicated domain.

On the other hand, AI as we know it remains ill suited for the complex, chaotic and disordered realms because computers, no matter how powerful or sophisticated, must have rules to follow. In my last post I referenced the AI assistant HAL from the film 2001: A Space Odyssey and the fact that the computer itself made the decision to kill the crew in as a way of resolving a conflict in programmed instructions. This is an example of an AI assistant attempting to solve a problem in a realm where it didn't belong.

While one message is that AI is inherently limited in realm, another message is that AI is potentially suited for realms beyond our present thinking. The primary factor in determining AI's usefulness is the availability and viability of knowledge, in other words, data. AI needs a lot of accurate data to perform effectively. 

The obvious place to start with AI is in the realm of known knowns. Here knowledge is the greatest and uncertainty the least. But don't stop there. Allow yourself to consider the known unknowns, those problems that can be solved through the sense-analyze-respond process provided sufficient data exists to perform the necessary analysis. Medical practitioners readily and confidently use AI to assist in their diagnoses and treatment plans, thanks to the ability of supercomputers to process millions of patient records in a matter of seconds, a feat the human brain cannot equal.

Lastly, when thinking about how AI might fit into your HR service delivery model, resist the temptation of focusing on hard-dollar benefits. AI may indeed pay for itself through headcount reductions, but remember that those reductions will be at least partly offset by new work associated with building, maintaining and generally supporting new AI tools. Instead, focus on ways in which bots can improve the employee experience by doing appropriate tasks better, faster and more conveniently. In particular, think about how AI can not only expand but improve the self-service experience for employees and managers. The payoff will come.

Should We Be Afraid of AI?

Were HAL, the space-age AI assistant from the 1968 film 2001: A Space Odyssey, human he would have celebrated his 50th birthday this year. That's really old in digital years. Yet five decades later HAL still epitomizes the computer that might someday take over the world. Could it happen?

               HELLO DAVE

               HELLO DAVE

To explore this question it's important to note that HAL stands for Heuristically programmed ALgorithmic computer -- the key word being "heuristically." A heuristic is a general concept, based on past experience, that can be applied to learn or understand something in the future. We use heuristics to discover or learn things for ourselves. In the film, the computer itself comes up with the idea of killing the crew members as a way to resolve the conflict between its general mission to relay accurate information, on one hand, and Mission Control's specific directive to conceal the true purpose of the mission from the crew on the other. HAL reckoned killing the crew would kill two birds with one stone. Pretty smart.

But today's cognitive computers don't work like HAL -- in fact, just the opposite. Unlike HAL, today's AI assistants aren't heuristically programmed. Instead, the basic AI model used today is based on a "deep learning" technique called backpropagation that was first published more than 30 years ago. Backpropagation is basically a process of working backward mathematically to evaluate and eliminate the universe of possible solutions to eventually arrive at the best one. If you're thinking this takes a lot of computing power you're getting the picture.  At the time the backpropagation technique was introduced, the computer powerful enough to run the algorithm had yet to be built. That changed eventually, and thanks to the tremendous processing speed of supercomputers like IBM's Watson and Google's DeepMind, deep learning machines eventually went mainstream.  

So, mathematically speaking, AI may as well be frozen in time. But computationally, due to supercomputers, it has moved light years ahead. IBM's Watson supercomputer, for example, runs 2,880 "processor threads" and has 16 terabytes of RAM, allowing it to process about 200 million pages of content per second. Not to downplay AI's mathematical sophistication, as a practical matter cognitive computing is more about brute computational strength.

AI supercomputers run a suite of cognitive software programs that allow them to see, hear, read, talk, taste, understand, interpret, learn and recommend, which is what makes them seem, well, intelligent. But because AI still relies on the mathematical technique of backpropagation, absent innovation in the underlying math, some predict progress in AI will level out and stall before nearing the fictional intelligence of HAL. Because while AI computers may get better and faster at processing data, they will forever be capable of just that -- processing data. In fact, AI purists argue that what we now call AI is merely machine learning, or the ability of the computer to learn from its mistakes and thus improve future accuracy. If they're correct, it's fair to say that the computing industry has decided to declare victory on AI by anointing machine learning as AI.

But even if machine learning falls short of being true artificial intelligence, this fact shouldn't diminish its great potential. To illustrate this point, consider the evolution of human flight. Early attempts at human flight were comical, if often tragic, versions of men bordering on insanity flapping mechanical wings to imitate the flying behavior of birds. Only later, thanks to the mathematical understanding of aerodynamics, was human flight finally achieved. But with the exception of these ubiquitous aerodynamic principles, humans don't actually fly like birds and never will. Nevertheless, human flying machines can do certain things far better than can birds.  Likewise, while cognitive computers may never work like the human brain, they can do certain things far better. Simply put, planes are like birds, but they're not birds, and computers are like brains, but they're not brains. Thus, unless a fundamentally different mathematical model for AI is discovered, AI can merely supplement but not replace human thought.

This is not to say that jobs aren't threatened by AI. Indeed, some tasks are just as prone to being assumed by bots as comparable factory floor tasks are to being automated. Clearly, people who perform such tasks should be fearful of AI's impact because jobs will go away. Fortunately, the loss of jobs will be, at least partially, offset by new jobs in the rapidly expanding AI industry. Whether the job losers will be able to migrate to these new jobs is yet to be seen, though history would suggest doubtful.

In my next blog I'll discuss a framework for assessing future opportunities for AI and the impact, negative and positive, on your workforce.