A few years ago, if you asked most technologists around Silicon Valley which technologies would affect the future of work, the discussion would have centered around AI, self-driving cars, fully digital assistants and the inevitability of long-term technological unemployment. Everyone was focused on the implications of AI as a fundamental, new prime intellectual mover in our world to replace the human mind. It’s like the disruptions experienced in the physical world when we shifted from wind and water power to coal and finally to gas and electric power.
In the last year, however, it feels like the future-of-work narrative has shifted toward more immediate and practical applications of technology that will be available soon—“near in technology”—to disrupt traditional work. There is an acknowledgment that machine learning and AI are powerful extensions of systems built on deterministic computing, not a fundamental disruption as a new, intellectual prime mover.
All of a sudden, a lot of companies are being formed where the key question is one of application and impact versus fundamental technology. Think of successful collaboration platforms like Zoom and Slack, the rise of robotic-process automation companies that came out of nowhere like UIpath and Automation Anywhere, and the reignition of billion-dollar labor marketplace businesses exemplified by Upwork and Scale.ai.
People interested in the future of work are now faced with an enormous wave of companies claiming that their product or technology fits into the narrative, and it is hard to sort out what really matters from noise.
This is the three-layer framework I apply to contextualize how knowledge work is being disrupted and what the biggest effects (and opportunities for company building) will be.
Layer 1: The Key Technologies of Work Disruption
A handful of fundamental technological shifts will disrupt the nature of work. Interestingly, most can be understood as extending frameworks and ideas that were historically applied to manufacturing, but now can be applied to knowledge work. As I see it, the key disruptions are as follows:
1) Dropping Cost of Measurement: You can’t systematically improve what you can’t measure, and where it has been relatively easy to measure linear step-by-step industrial production for a long time, it has historically been extremely hard to measure complicated knowledge work.
That is shifting rapidly. Inexpensive big data makes it possible to measure exactly what a salesperson, customer service agent, back-office administrative worker or lawyer was doing all day, for coaching and process optimization.
We are watching companies like Gong and Chorus start to log and transcribe the audio of every sales call and formalize the measurement of sales process. At Fin Analytics, a company I co-founded, we are building complete low-level clickstream oriented data sets of operations and customer service work, matching it to tasks from customer CRMs, and understanding the patterns of work and how to optimize people and process.
This, I believe, is the most fundamental disruption of work that is currently in play. Once work goes from a series of abstract boxes to a measured process that can be optimized, nearly everything else can change.
2) Consolidating Team Knowledge & Improving Standardization of Knowledge Work: Today, most organizational practices and standards are largely taught from person to person casually over time as embedded cultural organizational knowledge. It is only in the most structured work environments that day-to-day work processes are documented.
This, however, is changing. Once you can measure everyone’s process, it becomes possible to figure out the best process and then coach and standardize team practice around that. You can also hold people to process standards in ways that were historically impossible. This can make teams more efficient, make it easier to onboard new hires, and more (which we will discuss later).
To be clear, this doesn’t mean that everything is perfectly standardized and that human judgment doesn’t play a role anymore. But for most tasks and in most situations there is a better practice that can be followed. Where the best practice might have not been known before, technology now allows organizations to discover and formalize far more.
3) Improving Effectiveness of Automation: With improving standardization, it becomes possible to move toward the automation of some repetitive tasks. The entire narrative around RPA (Robotic Process Automation) fits into this key technological shift. More effective automation shifts the nature of human work away from repetitive tasks and toward more human-oriented tasks.
Automation ultimately has to be strategic. It only fits certain types of highly standardized tasks with limited branching logic, but it is without a doubt a key “output” of the formalization of work and workflows as much as it is an important piece of technology in its own right.
4) Dropping Communication Latency & Technical Cost: There will always be tasks that do not fit standardized or documented process. But next-generation communication tools make it easier and more efficient to deal with unstructured collaboration when necessary (although it is never optimal to need to communicate to complete a task).
Tools like Slack and Zoom fit into this part of the narrative, but so do task-management and project-organization tools like Trello, Airtable and Asana. As technology reduces the friction and cost associated with communication between an organization’s employees, it dramatically changes the patterns and nature of work.
To be sure, there is a tension between the drive toward more structured workflows and dropping communication costs. Lower communication costs in a vacuum actually encourage less structure and process, because any challenge or question can be easily discussed, and practices can be rapidly evolved. Ideally, organizations have the discipline to improve standardization and leverage dropping communication costs where they are most valuable. To be sure, it helps to measure and properly understand the full cost of communication in human time and attention—not just the now-low technical cost.
5) Improving Identity & Portable Reputation: Not everyone is equally good at all things. Better understanding of who people are (their identity) and their reputation (what others think of them) makes work far more efficient because it allows people to be slotted in and focused on the kind of work of which they are most capable. Of course, it matters greatly who “controls” this identity and reputation data. If companies control it, that may make them more efficient but not impact workers positively. If identity and reputation are more portable, then workers benefit but employers may not.
People forget how disruptive something as simple as LinkedIn was when it first became popular, and how even listing resumes publicly shifted the power dynamic from companies to people (at least at the high end of the employment spectrum). Portable identity and reputation, however, are only just really starting to have their full impact on the future of work as these systems get broader and touch more populations.
Layer 2: Business Practice Change Resulting from New Tools
If the first layer of change is all about how technology affects the nature of work, the second layer reflects the most important ways business practices are shifting because of technological change. There are four themes worth discussing.
1) The Move Toward Remote Intellectual Work: Since the dawn of the internet, people have been forecasting that intellectual work would move toward a more and more “remote” model, following the well-established trend of the globalization of physical production. It does finally seem like it is happening, because of the factors discussed above.
Dropping communication costs with products like Slack, Zoom, Github, Asana are—of course—a component of remote intellectual work seeming finally to be feasible. But it shouldn’t be lost on anyone that themes like improved identity and reputation frameworks, better work process and measurement is playing a huge role in the ability of people to source intellectual labor globally.
Identity and reputation is allowing more efficient sourcing and management of talent from anywhere. The standardization of work process is helping people efficiently work together at a distance by knowing their role and how to add value. Improved measurement is allowing teams and companies to properly judge and evaluate work without personal context and efficiently manage teams.
The promise of remote work is real and happening, not because of a single technology, but because of several technologies that together support a new format for organizing workers.
2) Move Toward Asynchronous Queue-Based Ranked “Piecemeal” Work: When process is ill- defined and work is not easy to package and route, work has to be completed as a single package synchronously, and frequently on a first-in-first-out basis. A team gets a task, and that team or group needs to complete that task or deliverable from start to finish before doing the next thing.
When work process is well defined, and the best person and timing for a task are well understood, then the synchronous and first-in-first-out requirement of many knowledge tasks can be dropped. Instead, the more efficient pattern of queue-based “ranked” work, where each person does the most critical or valuable task next, becomes possible.
Right now, most knowledge workers are given a handful of tasks and deadlines and told to work them to “completion.” That doesn’t make sense with modern technology. As standardization drops the cost of switching contexts and people’s skills are better understood, people’s focus and work should be continuously set and reset dynamically based on the needs of the organization overall.
3) Automation of Fully Standardized Tasks: It is impossible to ignore automation as a major implication of technology. Any process which is completely standardized and does not require judgment should simply be subsumed by deterministic computing. The story of the “human” computer and elevator operator will play out again, such that entire functions we think of as work today—like transcription, taxi dispatch or watching security cameras—we will no longer even think of as a possible human job category.
Beyond the simple replacement of labor, it is important to acknowledge that automation isn’t just about replacing entire jobs. Work tasks where some element of rote-repeated work is currently deeply coupled with judgment and decision making will have to be reconfigured so that human beings can focus on the parts of the work they do best, and machines can stamp out the rest. And further still, the types of services that require a lot of simple, repeated human work today will become far less expensive very quickly (which in many cases will boost demand for those types of services).
4) Extreme Variabilization of Labor: We all have experienced what variabilization feels like, using Uber and Lyft. Variabilization means that people not only will not have “careers,” they won’t even have jobs. Rather, the future is a world where we have reputations and identities, demonstrated abilities and skills, and knowledge. When we choose to convert our effort into money and/or work, there will be on-demand queues of available labor where our time and focus will be algorithmically prioritized. The increments of available work will be different for different people and types of work. For most it will be minutes or hours, for some it will be weeks or months. But very few types of work will last for years or more—outside, perhaps, of purely creative work or entrepreneurship.
Of course, the full variabilization of work isn’t going to happen all at once, but I do think it will happen more rapidly than people realize. That’s not because of some fundamental new AI-based abstract system, but simply because of the confluence of the above described relatively simplistic technologies.
Layer 3: Social Impact of Changing Business Practices
The first two layers of the shifting ecosystem of work we have discussed are practical and immediate. They are happening now. The third layer is in some ways the most important, but it is also the least clearly defined almost by definition. For the third layer you can follow the pattern of the first two layers of change into the world of social implications, but it is admittedly less clear how much of this change will play out. That said, here are the themes I think are most likely and most important:
1) Structural Shift of Knowledge-Power to Employers and Away from Employees: The standardization of tasks and centralization of knowledge of work process is obviously a huge benefit to employers and away from employees, so long as there is a larger pool of people who want to do the available work.
In factories, the process of production is sufficiently documented and understood by leadership that the people who do the job are labor. In most knowledge work contexts, however, the workers are not just the providers of labor, but also the store of all sorts of major and minor bits of institutional knowledge on how the work is done. As companies use technology to centralize knowledge and standardize process, the “knowledge” edge and value of employees is eroded, and the pool of people who can do a given task is expanded. There is no question this increases the leverage of employers and capital over labor.
Put differently, “knowledge” workers historically were paid for the combination of their knowledge and their work. In the future their knowledge will be centralized and standardized by companies, and while people might still be paid for their skill, it will be very hard for workers to own the knowledge they learned in school or accumulated on the job in any meaningful way.
This might seem fanciful, but the reality is it already is happening in even mundane places. Consider salespeople: Generations ago, they were paid not just for their selling skills, but also for their knowledge of their customers. In an effort to drive healthy efficiency, systems like CRMs centralize the knowledge of clients. That puts a premium on the craft of selling rather than any proprietary knowledge the salespeople have about the people they sell to.
The counter pressure on all of this has to be with expanding job opportunities. As remote work becomes more real, labor will be able to access the global demand for their work (not just the local demand).
2) Empowerment of high-skill over low-skill workers: Right now, the global labor forces that exist today mostly do very-low-skill manual labor following clear scripts and processes. Many of the currently globalized knowledge-jobs are going to be highly pressured by automation. It isn’t so much American workers who are going to have to deal with the shift of automation as workers in lower-cost countries who do a lot of outsourced work.
There is an argument, however, that as parts of those low-skilled jobs are automated, it actually creates an opportunity for higher-skilled, more expensive workers in places like the U.S. If a given task is currently 80% process-rote, and 20% cultural or judgment-based, the cost saving of offshoring outweighs the benefits of culture and judgment. As it becomes possible to automate the majority of the rote work, and the 20% of the culturally attuned work remains, it becomes efficient to hire higher-skilled workers at higher cost.
3) Employee services become self service: As knowledge work inextricably moves from lifelong careers to years-long jobs to “stints” and tasks, most employee services will need to become directly supported by the workers themselves rather than provided by institutions.
To be sure, the richest and most powerful institutions will weaponize benefits (as the Google and Facebooks do today). But for most when it comes to HR, ongoing education, career development, etc., which are today done by or paid by the organization, the expectation should be that these will become self-service benefits which workers invest in themselves.
The services will not go away. If anything, they might become even more important for workers to invest in. But they should largely become direct-to-consumer and aligned with the specific person rather than with organizations.
At the extreme, this theme broadly also encompasses labor power and organization. As the global labor supply expands and careers become jobs, workers will—at least in some forms—need to figure out how to efficiently “organize” in this new world.
4) Expansion of The Physical Landscape of Inequality: For all of human history, people have clustered around opportunity. As transportation has gotten less expensive and knowledge has rationalized bigger and bigger economies, we have experienced many challenging social implications of this reality—including things like the “hollowing out” of the middle of the country, and extreme inequality focused in cities like San Francisco.
Looking forward to a world where knowledge is centralized, communication is seamless, people are paid for skills but not knowledge, etc., we have to think hard about what is going to happen to where people live and the social landscapes of the country and the world.
One thing that might happen—and is anticipated by many science fiction writers—is that the wealthy cluster into just a handful of the most appealing places to live. Those places become impossible for everyone else to afford or access. Cities’ tax basis, and quality of life, will erode.
Another thing that could happen is that inequality outside of cities, in local towns across the country, starts to grow dramatically. Rather than the best and the brightest leaving their communities to go to cities, they stay in their local communities. This could have positive implications—especially if so much of the future economy has to be local-service-based. But it could also create tensions of its own as the extreme inequality we see in some cities becomes far broader across the whole country.
It is very hard to see which of these narratives will dominate, and the reality of what happens might come down to specific policy and tax decisions. But it is impossible to ignore as a major trend and implication of the future of work.
Conclusion
Imagine for a moment that all technological progress halted and we were stuck as a species with exactly the core technologies we have today for the next century. I would posit that even in such a world the landscape of knowledge work would change using existing technology to be nearly unrecognizable in the next 10 or 20 years.
That is what the core excitement is for many who see themselves as implementers versus foundational technologists. The tools have changed, which changes how knowledge work is organized and executed. The new approaches to work are going to profoundly change our world for the better, and profoundly challenge most of our assumptions about how we live and work today.
Many technologists I talk to always brush off the palpable angst that they hear from non-technologists about “robots taking their jobs.” I understand why. Almost everyone who understands the technology believes that people massively overestimate the core technologies of AI and machine learning (with the exception of a few PT Barnum types who revel in the controversy). That said, I think most technologists underappreciate just how disruptive what we already have built is going to be.
In the end, I am not a pessimist. I think that all this technology and change will open opportunity for people, expand the economy and create better jobs. But it will be very disruptive in the coming years.