AI reshapes jobs as firms battle data & governance

Snowflake has published new research on how generative AI and AI agents are affecting jobs, return on investment, and deployment hurdles. The findings are based on a survey of 2,050 business and technology leaders across 10 countries, including Canada.

The report paints a mixed picture of workforce change. It finds 77% of organisations report AI-driven job creation, while 46% report job loss. Among organisations that have experienced both hiring and role reductions, 69% say the net effect has been positive.

Canadian results show a similar pattern, with more organisations reporting both job creation and elimination in the past year. In Canada, 42% say AI-powered automation has both created and eliminated jobs over the past 12 months. Another 31% say it has created jobs. Smaller shares report job losses only (14%) or no effect (13%).

The research also points to growing interest in agentic AI, defined here as respondents' use of AI agents in production. In Canada, 31% say they are already using agentic AI in production, and another 32% say they are interested but still early in the adoption process.

Returns reported

Respondents reported positive financial outcomes from gen AI programmes, even as they described operational friction. Globally, organisations reported earning roughly $1.49 for every dollar invested in gen AI. In Canada, enterprises reported $1.45 for every dollar invested, up from $1.43 last year.

More than half of Canadian organisations said they have quantified a positive return. The report puts the figure at 53%.

The survey also suggests experience with AI correlates with confidence in its value. Among early adopters globally, 92% report positive ROI. Respondents also said organisations plan to allocate 22% of technology budgets to AI in the coming year.

Where jobs shift

The report breaks down where workforce effects appear most pronounced. Technical functions show both growth and contraction, depending on the task mix and level of automation. Globally, IT operations reported the largest job gains (56%), followed by cybersecurity (46%) and software development (38%).

Some of the same functions also reported the biggest reductions. IT operations again features prominently, with 40% reporting job losses. Customer service and support and data analytics each show 37% reporting job losses.

Overall, the findings point to restructuring rather than uniform expansion or contraction. Teams furthest along in AI deployment reported the most workforce change, with both gains and reductions as organisations reallocate work and redesign roles.

Data barriers

Despite reported ROI, organisations continue to cite major constraints on scaling AI. The report finds 96% of respondents face significant challenges, with data readiness and governance emerging as leading obstacles.

Nearly eight in 10 respondents reported technical or data-related challenges. Specific issues include breaking down AI data siloes (65%), measuring and monitoring AI data quality (62%), and preparing data to be AI-ready (62%).

Unstructured data readiness appears limited. Only 7% of respondents globally said more than half of their unstructured data is AI-ready. Canada scored above the global average, with 10% reporting that most of their unstructured data is AI-ready. India led on this measure at 14%, followed by Australia and New Zealand at 12%.

Governance and shadow use

The report highlights continued use of non-approved AI tools inside organisations. Globally, 57% of employees reported using non-approved AI tools, including 66% of C-level leaders. It also finds 60% of respondents say their organisations need greater investment in data infrastructure and monitoring software.

Canadian respondents reported lower use of unsanctioned gen AI tools than the global average. In Canada, 51% said they use unsanctioned tools, compared with 58% worldwide.

Enforcement of governance remains difficult in many organisations. Data governance was cited as "very challenging" to enforce by 22% of middle managers and individual contributors and by 19% of C-suite respondents.

Adoption patterns

Across functions, respondents reported the highest levels of active AI use in technical teams. Globally, 62% of IT operations teams reported active AI use, followed by data analytics (59%), cybersecurity (53%), and software development (50%).

Functions such as procurement, sales, and marketing were slower, with around 30% reporting active use. By industry, advertising and media led with 42% of organisations reporting AI in production, followed by healthcare and life sciences (34%). Manufacturing and technology each reported 32%.

In Canada, customer-facing use appears higher than the global average. The report finds 45% of Canadian organisations have put customer-facing gen AI in customers' hands, compared with 36% globally.

Respondents also reported extensive use of AI in software creation. The research estimates that about 48% of all code is AI-generated. It also finds 82% report improvements in code testing, bug detection, and resolution, and 80% cite gains in overall code quality.

Snowflake said the results show uneven workforce effects and highlight how much outcomes depend on how organisations apply AI in practice.

"AI's impact won't be uniform - some roles will dramatically amplify their influence and productivity, while others risk being left behind. The difference comes down to how effectively it's used: breaking down problems with first-principles thinking and guiding AI agents like high-performing teams," said Anahita Tafvizi, Chief Data Analytics Officer, Snowflake.

Omdia tied sustained returns to data management and operational controls as organisations move from pilots into wider deployment.

"The data shows that AI is delivering tangible returns, but scaling it successfully requires a strong data foundation and governance framework," said Adam DeMattia, Senior Director of Research, Omdia by Informa TechTarget.