The chief executive and chief economic opportunity officer of the world's largest professional network have published a new book arguing that AI transformation will be worker-led
Canadian HR leaders face a paradox that their counterparts in other countries do not quite share in the same form. On one hand, Statistics Canada data shows that the percentage of Canadian businesses using A.I. to produce goods or deliver services has doubled in a single year - from 6 per cent in the 2023–24 period to 12 per cent in 2024–25. Employee adoption has surged even faster: KPMG's 2025 Generative A.I. Adoption Index found that 51 per cent of Canadian employees are now using generative A.I. at work. Individually, Canadians are experimenting with these tools at pace.
On the other hand, the same KPMG research found that only two per cent of Canadian business leaders say their organisations are seeing a measurable return on their A.I. investments. In a global trust ranking conducted with the University of Melbourne, Canada placed 42nd out of 47 countries in trust in A.I. systems, and 44th in A.I. literacy and training.
Canada is a country where A.I. adoption is accelerating rapidly but A.I. value is barely materialising. That gap - between widespread use and virtually no measurable return - is precisely the problem that Ryan Roslansky and Aneesh Raman set out to address in their new book, published today.
What the LinkedIn executives are arguing
Open to Work: How to Get Ahead in the Age of AI, published by HarperCollins, is the work of Roslansky, CEO of LinkedIn, and Raman, the platform's chief economic opportunity officer. It draws on LinkedIn's global labour market data and case studies from early A.I. adopters to make a case that is simultaneously optimistic and urgent.
The book's most confronting argument for HR leaders concerns not technology but structure. "The org chart was built in the industrial age to bring order, predictability, and stability to rapidly growing organisations," Raman says. "Companies need to let that go, as it's going to hold back innovation."
That framing will resonate in Canada with particular force. A country whose largest employers include heavily regulated financial institutions, healthcare systems, energy companies, government agencies, and professional services firms - all characterised by deep hierarchies and well-defined reporting lines - faces a specific structural challenge. The org chart is not merely a diagram in Canadian enterprise; it is a compliance and accountability framework. Letting it go, or even loosening its grip, requires a different kind of institutional courage than it might in a Silicon Valley startup.
And yet the argument for doing so is compelling. "Where you're going to see the real returns on AI isn't just a new workflow around AI, but rather new work around human capability," Raman says.
The Canadian data problem: adoption without return
The 93 per cent of Canadian business leaders who told KPMG they are using or piloting A.I. technologies, matched against the two per cent seeing measurable returns, is one of the starkest numbers in recent Canadian business research. It is a portrait of a country that has invested heavily in A.I. experimentation and is not yet translating that experimentation into value.
Roslansky and Raman's book offers a framework for understanding why. Their argument is that A.I. productivity gains do not flow from deployment - they flow from the quality of human engagement with the technology. Workers who experiment intentionally, who develop judgment about when to trust A.I. outputs and when to override them, who use the time A.I. saves to do genuinely higher-value work - these workers generate returns. Workers who use A.I. superficially, without adequate literacy or organisational support, do not.
Read next: What's behind Canada's weak productivity growth?
The Canadian evidence supports this analysis. The KPMG index found that 83 per cent of Canadian employees who use generative A.I. want or need to upskill to use it more effectively - yet fewer than half say their organisation provides adequate support. IBM's 2025 research found that while 79 per cent of Canadian office workers say they use A.I. at work, only 25 per cent rely on enterprise-grade tools. The rest are using a mix of personal and employer tools, or entirely personal applications - what the industry calls "shadow A.I." - with all the data governance and compliance risks that implies.
This is not a technology problem. It is a workforce capability and governance problem. And it is, in essence, an HR problem.
The skills that survive - and what Canadian organisations are missing
Drawing on conversations with neuroscientists, organisational psychologists, behavioural economists, and talent leaders, Roslansky and Raman identify five human capabilities - the 5Cs - that A.I. cannot replace: curiosity, curage, creativity, compassion, and communication.
The framing matters for Canadian HR leaders because it challenges the direction in which most Canadian A.I. investment has flowed. Statistics Canada data shows that training existing staff is the most common operational response to A.I. adoption among Canadian businesses. But the nature of that training is critical. If it focuses predominantly on tool proficiency - how to use Copilot, how to prompt ChatGPT, how to navigate A.I.-assisted workflows - without developing the underlying human capabilities that make those tools genuinely productive, the return will remain elusive.
Read next: The career ladder is disappearing. Are taxes making it worse?
"AI can generate possibilities based on patterns. Humans decide which ones matter," the authors write of curiosity. On courage: "AI can calculate risk. Only humans decide what risk is worth taking." On creativity: "AI can remix what exists. Humans decide what's worth reimagining." On compassion: "AI can simulate concern. Only humans feel it and express it." On communication: "AI can translate language. Only humans can turn language into meaning."
For Canadian HR leaders designing capability frameworks, the 5Cs offer a lens that complements technical A.I. training rather than replacing it. The question is not whether employees can use A.I. tools, but whether they have the judgment, the courage, and the creative capacity to use them in ways that generate genuinely new value.
The skills imperative is urgent - and Canada is behind
LinkedIn data paints a picture of accelerating change that Canadian workforce planners need to internalise. The platform found that 24 per cent of the skills required for the average job changed between 2015 and 2022. Looking toward 2030 and factoring in A.I., the authors estimate that proportion could reach 70 per cent.
That number sits uncomfortably alongside Canada's current position in global A.I. literacy rankings. Forty-fourth out of 47 countries is not a position that supports the pace of change the LinkedIn data describes. The Conference Board of Canada's youth unemployment data adds a further dimension: Statistics Canada reported a youth unemployment rate of 14.6 per cent in July 2025, the highest since 2010. The workers most likely to drive A.I. transformation are entering a labour market that has not yet created the roles or the training infrastructure to bring them in effectively.
Nearly 90 per cent of C-suite leaders globally say accelerating A.I. adoption is now critical. Two-thirds say they will not consider job candidates without A.I. skills. Canadian HR leaders who are still building A.I. literacy programs on a three-to-five-year horizon need to reassess whether that timeline is compatible with the pace at which the job market is already moving.
The worker-led transition - and Canada's shadow A.I. problem
The book's most practically disruptive argument for HR departments is its insistence that A.I. transformation will not be driven by enterprise programs. It will be driven by workers experimenting in their daily roles, often without formal permission.
"It's going to be a worker-led transition, and so companies are going to have to figure out how to let individuals start to move into this new era in their day-to-day work," Raman says. "We have more autonomy than we often think in terms of pushing for what we want to do that might push our work to the next level."
Read next: Why Jack Dorsey's 40% AI layoffs are a wake-up call for HR
In Canada, this is not a theoretical prediction. It is already happening - and it is creating real risk. IBM's research found that shadow A.I. - the use of unsanctioned A.I. tools in the workplace - added nearly $308,000 per data breach in Canada in 2025. Only 39 per cent of Canadian organisations have workplace A.I. policies in place. Only 20 per cent of employees with access to workplace A.I. tools report receiving formal training. The majority are learning through experimentation, social media, and online forums.
The IBM data also found that 46 per cent of Canadian workers say they would leave their current employer for one that uses A.I. more effectively. The retention implications of lagging on A.I. capability development are not speculative - they are showing up in stated employee intentions right now.
For Canadian HR leaders, the worker-led framing creates a governance question that most A.I. policies have not yet resolved. If adoption is fastest among workers who experiment independently - using personal tools, crossing team boundaries, adapting processes without waiting for sign-off - how does the organisation create structured conditions for that experimentation while managing the data, privacy, and compliance risks that are particularly acute in Canada's regulated industries?
Canada's Personal Information Protection and Electronic Documents Act (PIPEDA) and the proposed Artificial Intelligence and Data Act (AIDA) create specific obligations around automated decision-making and the use of personal data in A.I. systems. For HR functions that are deploying A.I. in hiring, performance management, or workforce planning, understanding where these obligations apply is not optional. The organisations that build governance frameworks now will be significantly better positioned than those that address them under regulatory pressure.
The org chart question - specifically Canadian
Raman's argument that the industrial-age org chart will hold back A.I. innovation has a particular resonance in the Canadian context. Canada's most significant employers - the major banks, the healthcare system, the federal and provincial public service, the energy sector, the major professional services firms - are among the most hierarchically structured organisations in any advanced economy.
The value of that structure is real. It supports compliance, accountability, equity, and consistency in ways that matter enormously in regulated industries and public institutions. The argument is not that hierarchy is wrong, but that the specific logic of the org chart - rigid role definitions, vertical accountability, cross-functional barriers - actively prevents the kind of worker-led experimentation that generates A.I. value.
Raman urges executives to get comfortable with workers figuring out A.I. on their own, "even if those experiments cut across departments and job descriptions." For a Canadian bank, a hospital network, or a government department, that comfort will need to be very carefully scoped. But the instinct to lock everything down until a formal program is approved is precisely the instinct that the book argues will cost organisations the most.
What Canadian HR leaders should do now
The evidence from the book, combined with Canadian-specific data, points to four priorities.
Close the training gap before it closes options. Eighty-three per cent of Canadian employees who use generative A.I. say they need to improve their skills, but fewer than half feel their organisation provides adequate support. That is the gap where A.I. value goes to die. Role-specific, practical A.I. literacy training - not generic digital skills programs - is the most direct lever available to Canadian HR leaders right now.
Build A.I. governance before regulators build it for you. Canada is moving toward an A.I. regulatory environment through AIDA and evolving PIPEDA guidance. The HR functions most exposed are those using A.I. in hiring, performance assessment, and workforce planning. Getting ahead of that regulatory trajectory with documented governance, human oversight mechanisms, and employee disclosure practices is both a compliance imperative and a trust-building opportunity.
Create structured space for worker-led experimentation. The shadow A.I. data is a signal, not a threat. Employees are using A.I. without approval because the tools are useful and the formal channels are too slow. Designing safe, bounded experimentation frameworks - with clear data rules and feedback loops - is how organisations capture the innovation that workers are already generating, while managing the risks that currently sit entirely outside governance.
Develop the 5Cs alongside technical A.I. skills. Curiosity, courage, creativity, compassion, and communication are not soft skills that sit alongside A.I. capability - they are the human capabilities that determine whether A.I. capability produces value. Canadian workforce development programs that invest only in tool proficiency, without building these adjacent human capacities, are likely to remain in the 93 per cent that adopt A.I. without seeing a measurable return.
