Kathy Gibson reports – While artificial intelligence (AI) is being integrated into most types of work, there are signs it might be flagging.
This is among the headline findings from BCG X’s AI at Work report, which uncovered some unexpected trends as companies start integrating AI into their workforces.
Sylvain Duranton, global leader of BCG X, points out that AI has quickly become part of our working lives. Employees are demonstrating big improvements in positivity (52%), confidence (36%) and curiosity is 54%. Meanwhile, concern is down to 28%, with just 11% of respondents feeling anxiety.
But there are still gaps in AI adoption, which is stalled at 51% for frontline employees, compared to 52% last year. Managers, on the other hand, have increased from 64% last year to 78% this year, while leaders are at 85% adoption, from 88% last year.
An interesting finding is that the global south is still showing higher adoption of AI tools, with all of the survey countries in the south tracking at above 70%. With the exception of Spain at 78%, all countries in the global north have a lower-than-70% adoption rate.
“The levels overall are increasing,” Duranton says. “But the gaps remain.”
Workers still fear that AI will cause them to lose their jobs – but counter-intuitively, this is higher among managers and leaders (43%) than among frontline workers (36%).
This is a switch from last year’s results. “A lot has been said about whether we need keep humans in the loop when we deploy AI,” says Duranton. “Today, the question is: do we need to keep the boss in the loop?
“It poses the question about what the role of the manager is; and the value-add of management roles could be questioned by the deployment of AI in the workplace.”
The fear of being replaced is much higher in countries with higher usage. The Middle East, Spain and India are particularly high (63%, 61% and 49%). Duranton believes this could be because these countries have a high level of deployment and this recognise how much AI could do.
In South Africa, 28% of respondents fear losing their jobs to AI.
There is no doubt that AI has had a big impact of employees: 47% of them say the technology has saved them at least one hour per day. Most use this extra time to perform more tasks (54%); others finish work earlier and with better quality (52%), work on strategic tasks (44%), experiment with GenAI (36%) or pursue professional development (34%).
Organisations still face some major challenges in deploying AI, mostly boiling down to last of skills and training, poor access to tools, and a lack of leadership direction.
Vinciane Beauchene, global lead on human x AI, explains that only 36% of employees feel they have received proper training.
She adds that, to be effective, this should include at least five hours of instruction, in-person session and coaching are key components for effective training.
Five hours of instruction increases adoption by 19 percentage points; in-person training pushes it up by 12 percentage points, and coaching adds 14 percentage points. These interventions also increase employee confidence.
For organisations that don’t deploy AI, the risk is not so much that employees won’s use the tools, but that they will use unauthorised products – which is what 54% of employees say they would be if not given access to the right tools. And this number is higher for younger generation workers, at 62% of Gen-Z and Millennials, says Beauchene.
Management is notably lacking, with just 25% of employees experiencing positive leadership. “This is not only about the adoption of AI, but how positive employees feel and their positivity about GenAI on their career prospects.”
The bottom line for companies is about driving value, Beauchene adds.
Today, 72% of companies are deploying GenAI tools, but this only drives defused productivity, she explains.
Fifty percent of respondents are going beyond deployment to rethink end-to-end processes and functions, using GenAI to transform how you do your work. This drives much more value, removes bottlenecks and reinvents processes.
The third phase about inventing new business models and driving top-line impact. “It is about innovation fuelled by GenAI. But only 22% of respondents say their companies are doing this.”
Not surprisingly, the Reshape and Invent phases represent the biggest investment, but this is where the value of AI lies.
Financial services and technology are leading the pack when it comes to workflow design, both at 54%. Public sector is lagging at 31%.
The survey found that companies redesigning their workflows tend to invest more in people transformation – and it pays off. Offering proper training adds 18 percentage points to deployment effectiveness, providing support adds 19 percentage points, and tracking value-adds increases it by 23 percentage points.
As a result, employees say they save more time, are able to shift to strategic tasks, and believe that AI enables better decisions.
However, employees in companies reshaping workflows do feel more vulnerable to job loss, rising from one-third in companies that have only deployed AI to almost half in companies reshaping workflows.. “This reinforces the need for clear communication and proper upskilling,” Beauchene says. “Companies need to anticipate the impact on employees, transparently provide a vision, and build muscle to allow employees to evolve their careers.”
Agentic AI is getting a lot of attention, but so far only 13% of respondents see them being used in workflows.
And, while 77% of workers thing AI agents will be important in the next three to five years, only 33% say they have a good understanding of what agentic AI is and what is can achieve.
Nipun Kalra, leader of BCG X India, points out that most companies are still experimenting with agentic AI. Just 13% have integrated it into their broader workflows, 56% are using it experimentally in pilots or under human supervision, and 31% have not deployed it at all.
Brazil and India are leading in agentic AI deployment, at 18% and 17% respectively. South African adoption is on par with the UK and Germany at 13%.
Among the main concerns around agentic AI are a lack of human oversight (46%); unclear accountability when mistakes occur (35%); and the potential introduction of bias or unfair treatment (32%).
“When people are more familiar with AI agents, they see them as a valuable tool rather than a threat,” Kalra points out.
But there is a big gap – 46 percentage points – between these respondents and those who perceive agentic AI as a threat.
Duranton concludes that leaders need to push and support the deployment of AI, particularly on the training side; track the value they are generating with AI in terms of productivity, quality and employee satisfaction; invest in people to reshape workflows and unlock the value of AI; and experiment rigorously with agents to accelerate the experience curve, tracking impact and potential risk.