As new and emerging technologies like artificial intelligence (“AI”) become more prominent and prevalent in all aspects of society, this technology will play an increasingly important role in Extension’s efforts to connect with North Carolinians.

In order for Extension to remain the leading source of trusted, research-based information of the highest quality for our stakeholders, we must lay the groundwork for the strategic, successful use of AI in Extension.

Menu

• Overview
• Background and Context
• Best Practices and Considerations

AI Overview

Artificial Intelligence, or AI, refers to the intelligence of machines or software, as opposed to the intelligence of people or animals. AI involves computers, robots, software, etc. that have the ability to perform tasks associated with intelligent beings.

“Generative AI,” for example, refers specifically to digital tools – like OpenAI’s ChatGPT – that can generate a variety of content, including text, images, videos, music and more, based on data used to train them, along with ongoing user interactions and patterns.

• AI Overview Video (BBC)
• Using ChatGPT to Your Advantage (EIT)
• ChatGPT: Proceed With Caution (OIT)
• AI Resources for Higher-Ed and the Classroom (Provost’s Office)
• How AI Is Reshaping the Marketing Landscape (Poole College)

New AI tools launch by the day, with a variety of applications already available to assist with many Extension-related activities, whether it’s managing crop and soil treatments, translating video into other languages, optimizing personal nutrition, monitoring livestock health or crafting content.

Background and Context

Currently, there is no formal NC State or CALS policy regarding AI resources like ChatGPT. While AI is a complex and fluid topic, our goal is to share guidelines and best practices that facilitate productive, responsible use among employees.

• This is NOT a formal policy or regulation at this time
• This IS an attempt to provide initial guidance and best practices

We will pursue opportunities to collaborate, capture feedback and collectively update this resource. Over time, as the opportunities, issues and needs related to Extension are better understood, our direction could branch out in any number of ways.

• This is NOT comprehensive or final information
• This IS a living resource that will evolve and expand as we learn more

Back to Top

Best Practices and Considerations

While there are a growing number of AI resources available, we’ve placed initial emphasis on generative AI like ChatGPT, which is one of the most widely-used, accessible and potentially valuable tools for Extension.

Other prominent examples include Google’s Bard, Microsoft’s Bing Chat and visual platforms like OpenAI’s DALL-E 3, among many others.

Aside from some specific tips, however, these best practices can be applied to other AI as well. Above all, AI must be approached with respect, used responsibly, and balanced with judgment and principles.

Download ChatGPT Best Practices (Google Slides | PDF)

It’s Here to Stay

• AI is here to stay – we must learn to use it to our advantage
• How can we leverage AI to help augment our expertise/resources?

Be Mindful of Privacy

• AI tools like ChatGPT store every piece of info you share – don’t include anything you don’t want in the cloud!
• E.g. Personal ID details, client / stakeholder info, proprietary research, accounts, discriminatory language, etc.
   
  View OIT guidance on AI and sensitive university data

Not Just a Search Engine

• ChatGPT is a versatile tool that can assist with a variety of tasks (especially content generation)
• If you’re not sure whether it can help, give it a try!

You’re The Expert, Not the AI

• Don’t rely on the tool for subject-matter expertise in your area; plug in YOUR knowledge and let it help format or flesh it out
• Carefully vet ALL responses from ChatGPT and all AI for accuracy, language/tone, redundancy, appropriateness, etc.

Be Cautious of Bias and Inaccuracies

• Generative AI mimics humans’ online behavior, which is not always accurate, appropriate, etc.
• These tools draw from enormous datasets that often include bias, which can be further skewed by user patterns or “algorithmic bias” over time
• Again, review and vet all responses!

Engage and Experiment (Thoughtfully)

• Don’t be afraid to use it and experiment
• “Talk” to it and try different prompts to see what works best
• Be clear and specific with your queries
• Provide situational context and only give it one primary task at a time (break into steps if needed)

Write Effective Prompts

• When crafting a prompt, imagine you’re giving an assignment to an intern — be specific and consider three key components:
1. Background / Priming — Tell the AI a bit about the brand and/or your program, along with the role it should play. E.g., “You are a marketing manager creating content for (your brand’s) Facebook page. (Your brand) shares scientific knowledge from NC State University across North Carolina, providing people with practical solutions to improve their lives.”
2. Core Question — What do you want the AI to do (and to whom is it talking)? E.g., “Generate a social media post that tells young families in the Raleigh metro about (the brand’s new resource or event).” Add details as needed.
3. Output / Format — What do you want the output to look like? E.g., “This update should be a short social media post, less than 300 characters, and be formatted so it can be pasted directly into Facebook.”
• Remember that these tools are interactive, so if you don’t like your first response, revise and/or add more detail and try again.
• How to write ChatGPT prompts for the best results
• 7 advanced ChatGPT prompt-writing tips

Use it as a Draft Generator / Starting Point to Build on:

• Creating an outline for a letter, agenda, memo, etc.
• Drafting a message or social post using your info / data
• Creating a table
• Help with HTML for webpage
• View examples from EIT.

Be Responsible and Careful 

• Ongoing debates, issues and even legal proceedings regarding copyright, use of NIL, etc. (e.g. hard to know if copyrighted content is being pulled)
• Another evolving issue involves use of disclaimers on AI-generated content
• Did we mention to review and vet all responses before running with them?

Back to Top