Companies and their workforces are adjusting to a new reality: The “co-worker” helping them make better decisions or finesse a product on the factory floor isn’t necessarily a human. It’s a machine-learning program, a responsive robot or some other form of Artificial Intelligence (AI).
The rapid adoption of AI has triggered urgent calls for reskilling to prepare for the creation and eradication of jobs. But the majority of jobs will, in fact, be reconfigured as people and intelligent machines collaborate. So, before business leaders embark on a training revolution, they must first reimagine the very nature of work. The good news, according to Accenture research, is that 46 percent of corporate leaders think traditional job descriptions are obsolete. Twenty-nine percent say they have reconfigured jobs “extensively.”
The missing link: AI’s limits, people’s skills
Our research—which included on-the-ground ethnographic interviews in seven countries with people learning to work with intelligent machines—revealed the continued need for human intervention to help AI systems “learn” their tasks, to check the systems’ work and to keep the systems functioning properly. This reflects a view that AI applications are far from autonomous and, in some cases, cannot yet be fully trusted to deliver appropriate outputs. Some workers describe AI as “helping, but not doing.” Others point to AI performing well at single tasks but less so at more complex work. There is considerable input, adjustment and review needed, with humans responsible for the final decision-making and for taking AI-powered recommendations into the real world.
A production planning engineer at a Chinese auto company noted that while AI has enabled a ten-fold increase in the rate at which model iterations can be processed, that powerful capability is not sufficient:
Unlocking the value between human and machine
While 54 percent of employers in our survey acknowledge that getting human-machine collaboration right is critical to achieving their goals, few companies have worked out how to unlock the value that lies at the intersection of humans and machines. They must first explore how intelligent technologies affect jobs on factory floors and offices, from the customer-facing talent to the senior leaders.
Inevitably, automation will affect many low- and middle-skilled roles, so staying relevant in labor markets will require workers to uplift their skills significantly. The actual content of work is changing to such an extent that existing education and training practices will not be sufficient.
We suggest starting by first understanding the difference between new jobs that
will arise from AI and those that will be reconfigured.
Jobs that will be created: Nearly two-thirds (63 percent) of surveyed executives
said that intelligent technologies will drive job growth in their companies in the
next three years. AI is creating two major categories of jobs. The first is new kinds
of AI-oriented jobs, including roles in AI research and design, as well as
cybersecurity. The second is the use and maintenance of AI systems as employees
“train, explain and sustain” AI, a concept explored by Accenture’s Paul Daugherty
and Jim Wilson in their recently published book Human + Machine.
“In the insurance industry, companies are adopting intelligent solutions
specifically leveraging machine-learning technologies,” the CEO of a UK-based
insurance company told us. “We have introduced some new jobs to manage the
usage and operations of AI technologies.”
Jobs that will be reconfigured: More roles will be reconfigured than lost or
created as a result of AI. Intelligent machines can take on routine work while
helping people become more strategic and creative while engaging a greater
range of their skills. Operational jobs are likely to become more insight-driven,
while mono-skilled roles will become more multiskilled. (see Figure 1).
Consider how an aerospace designer today works labor intensively to create stronger, lighter airplane components using manual calculations. Through AI-assisted "generative design,” which mimics nature’s evolutionary approach to design, the same specialist can cycle through countless options, test configurations and nudge intelligent software to learn and improve with each iteration. Another example is the long-haul trucker who, with an autonomous vehicle, becomes an “in-cab systems manager,” performing high-level technical work such as monitoring systems and optimizing routes, potentially for multiple vehicles.
Our research of people in real-world applications also revealed that AI can increase levels of accountability. “Previously, superiors would place trust in our experience or our gut feeling,” said one of our interview subjects. “Now, insightful data prompts them to ask why we took an action when the AI told us we could have done something else.”
Steps to take
What are the practical steps forward? We recommend four key areas of focus:
Start with tasks, not jobs: Build from the bottom up, identifying tasks that are
critical to growth and to new customer experiences.
- Match these tasks to the resources best skilled to accomplish them: intelligent machines, employees, freelancers, supply chain partners or outsourcers;
- Then create a model that allows for this allocation to adapt as both
technology and market needs evolve.
How would that look in reality? As one Japanese banking executive explained, “Low-skill, back-end work is totally taken care of by intelligent machines. Other roles have been redesigned for shared responsibilities.”
Create new job descriptions: Break free from traditional functional roles:
- Create roles that are fluid and flexible, requiring a blend of skills that are
better suited to project work;
- Then orientate people to more valuable work that elevates the creativity,
human individuality and customer focus.
A trader at a Japanese investment firm told us: “We’ll get workers to become familiar with AI. They'll need experience as traders and must be strong in computers. They’ll need to understand how deep learning works, and that the data can't be useful or perfect without a knowledge of trading.”
Identify skills gaps: Only then should companies map new job descriptions to skills that exist within their workforce or in the wider ecosystem of partners:
- Where skills gaps exist, rapidly reskill existing employees or look for new sources of talent;
- Put in place longer-term training programs to create new skills that don’t yet exist in the labor market
One Indian telecoms company carefully analyzed changes in workflow to redefine
roles when intelligent technologies were added. As the company’s Chief Digital
Officer told us: “We then redesigned certain jobs, for example in customer
support and logistics support, and provided training to our employees to operate
these technologies in an efficient manner.”
Design for agility and open culture Because new work, roles and skills demand a more diverse and flexible workforce, companies should make wider changes to their organization:
- Break with today’s functions and processes to support the rapid assembly and disassembly of project teams, while recognizing the new forms of value that outsiders can bring;
- Give people greater autonomy and decision-making power as they shift from
functional roles to more collaborative teamwork;
- Inject new leadership capabilities to encourage experimentation, risk taking and customer focus.
As the Chief Digital Officer of an Indian telecommunications company told us,
“External people think out of the box, bring new ideas, and provide independent solutions to fit in our business environment.”