The broader question is to what degree they will enhance their skills with the help of automation - or eventually be replaced by it themselves. JL
Thomas Malone reports in the Wall Street Journal:
Regardless of what happens to the number of jobs, how is AI likely to change the structure of business hierarchies themselves? When AI does the routine tasks, much of the remaining nonroutine work is likely to be done in loose “adhocracies,” ever-shifting groups of people with the skills needed for whatever problems arise. The key for all companies will be to not assume any particular path. AI may create some more centralized hierarchies, and even more situations that call for flexible structures. The overall goal will be figuring out how to combine the different capabilities of people and computers
Ask people about artificial intelligence, and the discussion will most often turn to jobs: which ones will be eliminated and which ones will be created.
But regardless of what happens to the number of jobs, there’s another question that is less often discussed but crucial for maximizing both productivity and employee morale: How is AI likely to change the structure of business hierarchies themselves?
The obvious answer may be that the management structure is likely to get more centralized and rigid. After all, AI will help managers track more detailed data about everything their subordinates are doing, which should make it easier—and more inviting—to exercise stricter controls.
This will no doubt be true in some cases. But look more closely, and I believe the opposite is much more likely to happen in many cases. That’s because when AI does the routine tasks, much of the remaining nonroutine work is likely to be done in loose “adhocracies,” ever-shifting groups of people with the combinations of skills needed for whatever problems arise.
Consider that in a traditional hierarchy—such as a factory—large numbers of people are needed to do the core work of the organization: operating machines and doing the other tasks that machines can’t do. To coordinate the work of all these people, you usually need layers of managers in centralized hierarchies to make sure the people are doing their jobs according to a standard set of rules. In other words, a bureaucracy.
For example, imagine a traditional automobile factory without much automation. Executives and engineers decide on the product design. Then lots of people are needed in the factory to do all the routine work of cutting and painting metal, assembling parts and checking quality. Several levels of factory managers are needed to be sure the people are present, properly trained and doing their work the right way. This form of hierarchical management for large groups of people doing routine work has worked well for centuries.
But AI dramatically changes this picture. Robots already do many parts of the routine physical work that used to be done by people, and in the not-too-distant future, it’s plausible to imagine robots doing almost all of it. The job of the people is becoming primarily one of designing the new vehicles and production processes, repairing the machines when something goes wrong, and dealing with any other problems that occur during daily operations.
Worker freedom
It would be silly to try to manage the people who do this nonroutine work in the same way traditional factory workers were managed. Like workers in consulting firms, research organizations and innovative engineering groups, these nonroutine workers often know better than their managers what needs to be done and how to do it, so they need a great deal of freedom in how they do their work.
That could mean, for instance, that they need to work with a different group of colleagues to solve each different problem that arises. Fixing, say, a particular machine breakdown might involve electricians and mechanical engineers, while eliminating a frequent defect might require a materials scientist and an operations researcher. Any given employee might work on several projects at the same time, and the mix of projects would be constantly changing.
As long ago as 1970, futurist Alvin Toffler made a similar observation. “Far from fastening the grip of bureaucracy on civilization more tightly than before,” he said, “automation leads to its overthrow.” And Toffler popularized the term “adhocracies” for the looser, less hierarchical organizations that are likely to replace many hierarchies.
In a sense, AI is just one more wave of the kinds of automation Toffler was describing. In online retailing, for example, computers already do most of the actual selling, AI algorithms will increasingly automate pricing decisions, and robots will do much of the shipping. Or take accountants preparing corporate tax returns: The numerical calculations are already automated, and more and more of the judgments about how to apply tax law will be automated over time. And in the mortgage-approval process, most of the work that is done by humans today is likely to be automated with AI and other computational tools in the future.
In all these cases, we should expect to see the people who do the remaining nonroutine work organized in ways that are more like adhocracies than traditional hierarchies. For instance, the people in an online retail company might be mostly self-organized in temporary teams that come together to make all the software and other changes to introduce completely new types of products or promotions. An accounting firm might let the human accountants form their own shifting teams to analyze new tax regulations and the software and marketing changes needed to accommodate them. And a bank might have similar project teams to create new mortgage products, together with the associated software, marketing and underwriting methods.
In the details
None of this is to say that all jobs will become nonroutine and less bureaucratic. In fact, most of today’s jobs include some tasks that will be automated in the foreseeable future and some that won’t. For instance, even tasks that seem routine to human housekeepers, like straightening up loose items in a bedroom, are still too complicated for robots. And in the discovery phase of lawsuits, law-firm associates sometimes scan large numbers of documents for keywords—a routine process that computers can do much better than people. But even when the keyword searching is automated, deciding whether a specific document is relevant for the case at hand will often still require human common sense and legal reasoning.
In cases like these, the organizational changes that result from introducing AI will depend on the details of the situation. For example, if insurance companies let AI systems approve all routine claims automatically and send only the highly nonroutine claims to humans, then the human claims adjusters will probably work in organizations that are more like adhocracies. Handling each different claim, for instance, might require a different, temporary team with experts for the specific types of losses and locations involved.
By contrast, a high-end clothing store chain might automate checkout and delivery while still using lots of in-store salespeople to provide highly customized service. Computers aren’t even close to having the social skills that good human salespeople do, but these skills are fairly routine for the people who have them, and thus the chain might still use a fairly traditional hierarchy to manage the salespeople in all its branches. In fact, a company like this might even try to create a more centralized hierarchy by using various kinds of AI tools and video cameras to observe, evaluate and coach the salespeople as they do their work.
The key for all companies will be to not assume any particular path. AI may create some more centralized hierarchies, and even more situations that call for flexible structures. The overall goal, though, will likely always be the same: figuring out how to combine the different capabilities of people and computers into “superminds” that are smarter than anything we’ve ever had before.
0 comments:
Post a Comment