Working Smarter with AI

Working smarter with AI: what it means for Extension

Artificial intelligence is everywhere lately--every podcast and newspaper is exploring what it means for the economy, the environment, the workers and professionals, and the human race at large. For some of us, it's exciting. For others, it feels overwhelming, or even threatening. I explored working with AI and what it means for Extension at a recent talk at the All Extension Conference in Duluth. This post is a summary of some of my thoughts and an invitation to continue this discussion with your colleagues.

In ELT, our goal is to explore how AI can support the great work you (an Extension professional) already do. We don't want to add work to your plate or have us all chasing the latest tech fad. But there are ways AI can potentially help us work smarter, not harder. 

I find it helpful (comforting?) to remember how the web transformed how we help people, without changing why we help them. 

Extension's work still matters

Some reports predict that educators are among the professions most at risk from AI (source). That can sound  rather... unsettling, to say the least. But predictions like this focus on job descriptions and task lists, not the human connections and trust that are at the heart of good Extension work.

Extension work is needed in the age of AI

• Extension’s work provides human connection
• Trust and integrity are at the core of Extension’s work
• Every community is unique, Extension is present
• Extension bridges the gap between technology and human needs

(Source: pulled from the live webinar Artificial Intelligence Use in Extension: Insights from Ohio)

Where AI can help (and where it can’t)

A useful way to think about AI is along a spectrum of tasks. We all wear many, many hats and do a huge variety of tasks, some of which are probably not worthy of our professional skills. I've organized common Extension tasks along a spectrum:

Highly AI suitable tasks

These tasks are where AI can give you a solid start and save significant time.

• Format conversions (e.g. a report to a blurb)
• Research synthesis across multiple sources
• Brainstorming and ideation

Medium AI suitable tasks 

These tasks are where AI may be helpful in some aspects, but the content, context, sourcing, and verification need to be done by humans.

• Initial content scaffolds, when scoped with intent
• Audience-specific messaging (tailoring)
• Program evaluation questions
• Grant writing (where allowable)

Low AI suitable tasks (i.e. uniquely human)

These tasks are ones where AI simply can't replace human judgement and relationship skills.

• Relationship building and maintenance
• Reading the room dynamics and adjusting on the fly
• Conflict resolution and sensitive conversations
• Community needs assessment through observation
• Building trust and credibility

Extension's AI paradox

Here's the paradox: AI is really good at tasks that look like education (processing information, creating content, explaining concepts). But real learning happens in the space between people, and it's more about trust, relationships, and meeting people where they are. The question isn't whether AI can do some educational tasks (it usually can); it's how we use it to amplify the other parts, the uniquely human contributions to education. We need to be careful in our AI use to keep our authenticity, credibility, and human connection.

Guardrails and principles

With the above in mind, Extension has adopted principles to guide AI use:

• Protect sensitive information
• Maintain credibility
• Credit AI when it contributes
• Respect environmental impact (AI uses significant energy)
• Commit to continuous learning

These principles ensure AI supports our mission rather than undermines it. More information on these principles and how they were developed is available on our Extension Intranet.

Additionally, the University of Minnesota has approved specific AI tools (Google Gemini, Google NotebookLM, and Microsoft CoPilot) for safe and responsible use. (Tools like ChatGPT are not secure for UMN data.)

Practical ways to use AI

Here are three common Extension scenarios where AI can help you work smarter.

1. From research to usable content: summarize dense research papers for specific audiences, then build outlines for workshops or newsletters
2. Tailoring communication: rewrite a single message for different audiences, such as government officials, property owners, and youth.
3. Program management: brainstorm volunteer roles, draft recruitment blurbs, and identify potential event needs

In each case, AI helps overcome the "blank page" problem. Extension professionals still bring expertise, judgement, and context that will make the end product truly effective. 

We have more ideas on the Generative AI ideas & examples Intranet page, and continue to add more.

What now

As most of us can attest, there is a learning curve to using generative AI. It takes time to experiment before it starts saving time and being worth the effort. Once you become comfortable with where AI can help and where it can't, AI can save you time on routine tasks and allow you more time to focus on the uniquely human tasks on your plate.

Nobody has a perfect playbook for using AI. What we can do in Extension is build our literacy, experiment thoughtfully, and explore how these tools can support the human heart of Extension work. So keep exploring and share what works with your colleagues. Try out the approved tools and see which ones you prefer for what kinds of tasks. The best ideas for working smarter with AI will come from all of us, together. Happy exploring!

Article by Amy Baker, Extension Learning Technologies, amy@umn.edu.