Most conversations about artificial intelligence (AI) start with which tool to buy. I think they should start somewhere else.
AI is a leadership question before it is a technology question. It raises the same questions I have studied for years: who is in the room, whose data and voices count, and who is accountable for the decision.
My approach is human-centred, AI-powered and built around a “human in the loop.” AI helps with drafting, research and administrative tasks, so people have more time for thinking and relationships. A person stays responsible for every output. I also want to be honest about what we do not know. We are still learning, and the long-term effects of AI on work, learning and relationships are not yet clear. That is a reason for care, not a reason to look away.
Over the past few weeks, I have been in two rooms that sharpened my thinking: two days at the Human + AI Leadership Lab at Skills for Change with my Cohort 1 colleagues, and a fireside chat with the Junior League of Toronto. Four questions came up again and again.
1. Who is in the room?
Women are about 22% less likely than men to use generative AI, according to a Harvard Business School review of 76 sources. The International Labour Organization (ILO) found that 29% of female-dominated occupations are exposed to generative AI, compared with 16% of male-dominated ones. The jobs women are most likely to hold are the ones AI is most likely to change.
The gap in use follows a gap in support. In a Robert Half Canada survey reported by Benefits Canada, 85% of men felt confident using AI at work, compared with 72% of women. Formal training reached 38% of men and 29% of women, and 28% of men were given work time to experiment, compared with 21% of women.
This matters beyond fairness. AI learns from the data and questions it receives. When fewer women use it, it learns less about what women need and ask for, which makes it less useful to us. Women need to be shaping these tools, not watching from the side.
What helps:
Form an AI committee with people from different roles, levels and identities, much like an employee resource group (ERG).
Let that committee plan rollouts, build training and act as peer educators.
Protect paid learning time during work hours.
At the Lab, Heather Watts, Principal of First Peoples Group, led a session called Grounded in Connection. It invited us to think about technology through Indigenous ways of knowing, including the Two Row Wampum and the Dish With One Spoon. It was a reminder that technology is not neutral simply because it is new.
2. Whose data counts, and who owns it?
AI amplifies what we feed it. I call the sources of information feeding your AI tools your “data rivers” (i.e. internal documents, your customer database, public web content and vendor data). Map them, and name an owner for each.
The stakes are highest in hiring. Stanford researchers studied four million applications screened by one vendor's algorithm and found that 26% of applications from Black applicants and 15% from Asian applicants went to positions where the tool showed adverse impact. Adverse impact means a group is recommended at less than 80% of the rate of the most-favoured group. The paper also notes that more than 60% of the Fortune 100 use another vendor's algorithms, so when many employers lean on the same few tools, one tool's bias can follow a candidate from company to company. My colleague Nooreen Rahemtullah breaks down the research in We Read the Stanford Study on AI Hiring So You Don't Have To.
What helps:
Audit the data going in, and keep a person checking every output.
Know what should never go in. Confidential client information stays out. For First Nations data, the OCAP® principles (Ownership, Control, Access and Possession) set the standard. OCAP® is specific to First Nations, and the CARE principles (Collective Benefit, Authority to Control, Responsibility and Ethics) speak to Indigenous data more broadly.
If an Elder shares feedback or teachings, they do not go into a chatbot.
3. What are our guardrails?
Without approved tools and a clear policy, people use personal accounts for work. This is sometimes called shadow AI, and it puts organizations and clients at risk. McKinsey's State of Organizations 2026 survey of more than 10,000 leaders names change management and regulatory, ethical and legal risk among the top barriers to AI adoption. Those are governance gaps.
What helps:
Provide approved tools in one place.
Write a short AI policy and explain the reasons behind it.
Tell clients how you use AI, and where you do not. We publish ours here:
Guardrails come before speed. They are also what make speed possible, because people use tools more confidently when they know the rules.
4. Who is accountable when it is wrong?
Human in the loop only means something when a named person is responsible. AI can produce a answer that sounds confident but is not based on fact, often called a hallucination. If a tool cites a report, read the report.
Accountability is also about trust. Only 34% of Canadians trust companies that use AI, according to the TELUS 2026 AI Trust Atlas. Organizations earn trust by being clear about who reviews what, and by reviewing their own policies regularly. Because we still do not know the long-term effects of AI, I review my own policies every six months.
Where to start this quarter
Leaders have to walk the walk. If you hand AI to someone else, you cannot set direction or notice what is missing. Here is a starting plan for human resources (HR), diversity, equity and inclusion (DEI), and executive leaders.
Month 1, clarity: name an owner for your data and AI outputs, and list every AI tool in use, including personal accounts.
Month 2, capability: protect paid learning time, launch your AI committee, and block 15 minutes a day to learn one tool yourself.
Month 3, accountability: publish a one-page AI policy, audit one process or data source for bias, and track who has access to training.
If you are just getting started, try this: pick one real task you already do, use AI on it for 15 minutes, and compare the result with your own expertise.
Women need more than confidence. They need better systems, and leaders willing to build them.
Let Us Support You
Curated Leadership runs a workshop on ethical AI use that helps teams build literacy, set guardrails and keep people at the centre. For HR, DEI and executive leaders at large employers ready to lead AI adoption inclusively, email hello@curatedleadership.com or book a discovery call. You can also connect with me on LinkedIn or learn more about me.
Listen to the Curated Conversations podcast for more on inclusion and belonging.
