How Different Types of AI Work

A Nation's administration team needs to write a new housing policy. It's a big job pulling together legal language, community priorities, past policies, and the voices of the community. The team is small and stretched thin.

They try a generative AI tool. They feed it existing background documents, a summary of what the new policy needs to achieve, and examples of how their other policies are written. The AI produces a working draft in minutes.

From there, the legal team reviews the draft for compliance. Leadership reads it against their governance principles. Elders check that cultural protocols are respected. The community gets a say on priorities. The final policy is shaped by people, but the AI saved weeks of starting from scratch.

The AI wrote a draft. The community made a policy.

Generated content always needs a human to check it for accuracy, bias, and cultural safety. The AI doesn't know what's true. It knows what sounds plausible.

When to be cautious or avoid GenAI

• Never input private, sacred, or sensitive personal or community knowledge. This includes being cautious of generating text, images, or videos that depict your people, your ceremonies, your language, your regalia, etc.
• Never use AI for medical, financial, or legal advice.
• Always review output for bias, hallucination, and cultural appropriateness before use.

A 2026 study by the Centre for International Forestry Research (CIFOR-ICRAF) found that AI models used to predict and map wildfires perform significantly better when they incorporate Traditional and Indigenous Knowledge (TIK). Indigenous communities hold generations of place-based understanding about fire behaviour, seasonal patterns, land use, and cultural burning practices. This is knowledge that typical AI systems, which rely on quantitative data alone, consistently miss. Rather than replacing that knowledge, the study found AI works best as a complement to it: AI could benefit from TIK through better place-based contextualization, while TIK could benefit from AI through the archiving and translation of knowledge. If a community chooses to participate in such efforts.The benefits, however, only materialize under specific conditions. The study is clear that AI-based fire management systems must be developed in ways that respect Indigenous governance, participation, and data sovereignty. This means Indigenous communities holding decision-making authority from the start, formal agreements establishing community ownership and control over their data and any AI systems that use it, and capacity building that creates local technical expertise rather than long-term dependence on outside institutions. Frameworks like OCAP® and the CARE principles provide the governance foundation for this kind of partnership. Without those conditions, the study warns, integrating TIK into AI systems risks repeating a familiar pattern: a long history of extractive engagement that strips Indigenous knowledge of its cultural context, harvesting it for scientific or commercial purposes while leaving communities with neither sovereignty nor benefit.

The band administration office handles a mountain of work: council meeting agendas, permit requests, policy documents, grant correspondence, Elders program coordination, and more. Two admin staff are responsible for all of it.

An assistive AI tool is brought in. Documents, schedules, and past correspondence are uploaded into the system. Within a few days, the AI is:

• Automatically organising incoming files by department and topic
• Drafting responses to common permit requests based on approved templates
• Generating meeting agendas from calendar entries and past minutes
• Flagging upcoming deadlines for grants and reporting

The admin staff review and approve everything before it goes out. But instead of spending their days on filing and formatting, they're spending more time with community members who need support. The work that matters most gets the time it deserves.

Community members often have questions about services outside of office hours — "How do I apply for housing support?" "When does the youth program run?" "Who do I call about an urgent matter?" In the past, those questions went unanswered until Monday morning.

A community-controlled chatbot is built using Nation-approved information — the kind of answers the office staff would give. Community members can now ask questions through the website at any time and get clear, step-by-step guidance. Complex or sensitive situations are always referred to the staff.

Importantly, in this example, the Nation controls what the chatbot knows and how it responds. It doesn't pull from the internet; it only draws from approved information.

Each morning, the lands team begins their day by checking in on the territory. For generations, this has meant walking the land, observing the water, and the patterns of the animals. Today, that knowledge is still central, but it is now supported by new tools.

Across the Nation’s territory, water sensors quietly collect data, satellite images track changes in vegetation, and land guardians submit field notes from their time on the land. There is more information than any one person could reasonably review each day. This is where an agentic AI system comes in.

Behind the scenes, the AI agent continuously reviews incoming data from these different sources. It notices when something shifts—a subtle change in water quality, an unusual pattern in wildlife movement, or a trend that has been building over weeks. It organizes this information and prepares a simple summary for the lands team. When the team logs in, they don’t just see raw data; they see a clear picture of what needs attention. The system might flag, “Water quality readings in this area have dropped over the past three days,” or “Wildlife activity appears lower than usual in this region.” It may even suggest that a site visit could be helpful. But the AI does not interpret what this means for the Nation. That responsibility remains with the people.

Lands staff and knowledge holders review the information, bringing their understanding of the territory, the seasons, and their relationships with the land. They decide whether the change is concerning, what actions to take, and how to communicate it to leadership and the community.