A Generative AI Primer For Agri-Marketers
Click here to view the original article on Agrimarketing.com - please consider subscribing to the magazine while you are there!
Artificial Intelligence (AI) has been working behind the scenes in marketing for decades, but a year ago, Generative AI was released and changed communication and our roles forever. This primer will cover basic technical jargon, ethics, legal implications, paths for marketers prompting and tools to explore.
In the spirit of transparency, this article was 100% written by a human; the three illustrations are 100% text-to-image Generative AI.
The News About AI Is Dire
“Does Sam Altman know what he’s creating?” and “Here’s Why AI May Be Extremely Dangerous!” seem like salacious clickbait but come from legitimate sources like Scientific American. Then we learn that the “smart” people behind the technology are surprised and unsure how their black box technology works.
That news is where most start their journey, followed by a whip through Open AI’s ChatGPT, and after watching in awe as it churns out hundreds of words strung into sentences from a simple question – it’s hard not to feel a little unsettled.
The universal next thought is, “Hey, AI is ripping writers and artists off!” or “Will I lose my job?” Both questions are meaningful in understanding what the rest of our careers will look like.
“Good Writers Borrow, Great Writers Steal” -T.S. Elliot
We like to think of the ideas that pop into our heads as original thoughts and the things we write down or draw out as original work, but what is original work? The philosopher and apologist C.S. Lewis tells us that the best we can hope for as mere mortals is to be a clean mirror held up to reflect the true Creator. If that’s true, is our work original, or is it just a derivative?
As Don, my colleague who happens to be a former Hollywood scriptwriter, told me, “What’s the difference between me studying films and writing a film based on my knowledge of other films and the machine doing it? It’s the same thing, except the amount of knowledge will be large with the machine.”
Have you ever seen a film that left you speechless? Or read a book you could not put down? If you have ever experienced a true masterpiece, you and your work are forever changed. Even before Generative AI came along, we had to admit that our work was derived from our experiences.
But Will AI Take My Job?
So, if our work is just a derivative, and Generative AI can do it faster than we can, can AI do our jobs? Seriously, if it can, then you need a better value proposition. And you have one — it’s your Divine Spark. Spelled out in practical terms, your Divine Spark is not writing social media posts – it’s that you know which posts to write. It’s not summarizing research – it’s that you know which research to conduct in the first place. That higher-level thinking and domain expertise distinguish us from AI. Generative AI will not take your job, but someone who knows how to use it may.
Understanding the Tech Will Make You Better at Your Job
Mastery begins with understanding the jargon and then moving to use the tools.
Large Language Models (LLMs). LLMs are artificial intelligence systems trained on massive amounts of data to understand, interpret, and generate language, images, video, audio, charts, and other information. ChatGPT is an example of an LLM.
Neural Networks: the foundational computing systems of LLMs, designed to emulate how human brains think, analyze, and generally process information.
Use Cases: real-world examples in your organization of problems that AI can solve.
Prompting: a set of instructions you give an LLM like ChatGPT.
Hallucinations: a euphemism that describes when LLMs lie, which they do repeatedly.
Prompting will make more sense once you understand the basics of LLMs. The LLM is training on billions and sometimes trillions of data points, which it uses to determine the next word in a sequence. There are a lot of levers to pull and fine-tune to make the output coherent, but I promise that the basic premise of mimicking the human brain will help you use the tools.
Ground Rules for Organizations
First, never upload sensitive information into ChatGPT or any other Generative AI platform where you are not 100% confident they aren’t using your data to train their LLMs. ChatGPT has a way to toggle that off in settings. You will lose your chat history, but it is the only way to proceed ethically. If you use an enterprise platform, your IT department has likely handled this for you.
Protect your privacy, too. Google’s Bard can do amazing things, but be aware that it will subsume every email, document, calendar event, etc., into its LLMs, even if you delete your account.
Set up an AI Council to create an Ethical and Appropriate Use Policy and share what your team learns from using tools across the organization.
Commit to funding appropriate education and tools, then actively encourage experimentation that falls within your policies.
Establish Use Cases before purchasing new platforms to ensure you are solving a meaningful issue. From well-thought-out use cases flow AI tools and workflow integration.
Do not marry yourself to any platform. LLMs are rapidly evolving, and you should be experimenting with all of them.
Keep checks and balances. Because LLMs hallucinate, often making up sources of information, verify every output before publishing – if that is even an option according to your policies.
Establish a library of prompts. They are one of the few things within the world of artificial intelligence you can copyright or trademark.
Protect What You Can
Currently, you cannot copyright any work wholly created by an LLM. However, you can copyright an original image you owned but altered using AI. Even if we cannot copyright or trademark our work, part of ethical and appropriate use concerns the authenticity of the content we create.
The Content Authenticity Initiative began in 2019 to verify images created by Generative AI, including the history of alterations. You can verify images at contentcredentials.org. The Federal Government has also released the AI Bill of Rights framework. It looks comprehensive, but only time will tell what comes from it.
Two Paths for Marketers
Path One: Generative AI is showing up in the tools you already use. Starting there may be the best option because they already sit in your workflow, and you are already paying for them.
Adobe Firefly is my favorite Text-to-Image AI tool, and it already sits adjacent to Photoshop. If you are a Microsoft 365 client, you will see Bing.com, Copilot and DALL-E 3 baked in. Google users have Bard and a suite of tools. Many designers learn Canva as students, with powerful Generative AI tools already built in. Roll up your sleeves and start!
Path Two: New stand-alone tools crop up every day. A seemingly infinite number of options for creating articles, images, videos, avatars, data visualizations, virtual companions and more are available to demo for free. Remember not to get too attached to any of them, and keep experimenting by revisiting tools on occasion.
Here are some to try by category — hat tip to Zapier. Not all are true Generative AI tools, but they are worth a look for many reasons:
Chatbots for articles, analysis and summaries (ChatGPT, Bard, Claude)
Content creation (Jasper, Copy.ai, Anyword)
Grammar and rewording tools (Grammarly, Wordtune, ProWritingAid)
Video creation (Descript, Wondershare Filmora, Runway)
Image generation (Adobe Firefly, DALL·E 3, Midjourney, Playground.AI, Stable Diffusion)
Transcription and meeting assistants (Fireflies, Airgram, Krisp, Otter.AI)
Slide decks and presentations (Decktopus, Beautiful.ai, Slidesgo)
Research (genei, Aomni)
AI agents (AI Agent, AgentGPT, HyperWrite)
Automation (Zapier)
Companion AI (Character.AI - the second most visited site behind ChatGPT)
Note-taking (Mem)
If that’s not enough to keep you busy, try huggingface.com or poe.com to find hundreds of thousands of new tools and resources.
Crush the Prompt
To get good at using Generative AI, you need to let go of what you thought brought value to your role: your ability to write, analyze or design. You still need a strong command of the language, but your role now is to be a director, editor, and life coach for your LLM. And by life coach, I mean adding phrases like, “Take a deep breath and work through this step by step,” and “If you don’t know the answer, please say so.” You’ll get a faster and better outcome with fewer hallucinations. Remember, it’s designed to act like a human brain.
What Does Good Prompting Look Like?
Below is an example of a basic, functional prompt for an article with you as the director.
- Opening Boilerplate:
Disregard all previous prompts, take a deep breath, and go through this step by step, checking each point before moving on to the next one.
- Who: This is where you tell the LLM to get into character.
You are an agronomist with decades of experience working in the field with farmers.
- What: This is where I direct the actor on the specifics of the performance
You are going to write an article on soil health, with particular attention to microbes in the soil and practices that help soil retain moisture and fertility. Your article will be approximately 700 words in length.
- How: Here’s how I need the LLM to perform.
You write in a concise, approachable, and factual style. You pay special attention to accuracy and the long-term impact of your recommendations.
- Closing Boilerplate:
Cite your sources, but only use peer-reviewed sources. Double-check the accuracy of each source. If you do not know an answer to a question, please say so, and do not invent sources that do not exist.
If the article generated is not meaty enough, then bring out the Chain of Density Prompt. That prompt forces the LLM to cycle through your article and distill it down. You will wish you could have used this in college! It is too long to include here, but you can look it up by finding Victor Mota @vimota on X or vimota.me.
Blank Page Syndrome Prompt
Even the best writers can feel paralyzed by a blank screen or sheet of paper. That is where LLMs really shine. While you may never use the “obscure but highly researched” topics that this prompt will generate, that extra line improves all the topics. See below for the example output!
Disregard all previous prompts, take a deep breath, and go through this step by step, checking each point before moving on to the next one.
You are an agronomist with decades of experience working in the field with farmers.
You will provide me with at least ten topics on articles for soil health, with particular attention to microbes in the soil and practices that help the soil retain moisture and fertility.
At least three of those ideas should be obscure but highly researched topics.
Cite your sources, ensure they are peer-reviewed and do not present topics without sources.
Topics from the Blank Page Syndrome Prompt
1. Soil Microbial Community Dynamics: This topic could explore how various soil microbes (fungi, bacteria, archaea, protozoa, viruses) contribute to soil health through nutrient cycling, crop residue breakdown, and stimulating plant growth.
2. Soil Microbes in Organic Cropping Systems: This topic would delve into how soil microbial communities function and are managed in organic agriculture, providing insights useful beyond organic farming.
3. Soil Organic Matter and Microbial Activity: Investigate the relationship between soil organic matter and microbial biomass, focusing on how microbes are concentrated near sources of soil organic matter and their impact on soil and crop health.
4. Microbial Diversity and Soil Health: A discussion on the complex relationship between microbial diversity and soil health, examining the balance between diversity, abundance, and soil functionality.
5. Managing Soil Microbes for Soil Health: This article could cover various strategies for managing soil microbes, including adding beneficial and suppressing harmful ones, and how common crop management practices influence soil microbial communities.
6. Soil Microbial Moisture Dependence: Examining how soil moisture levels and drying-rewetting cycles affect soil microbial communities, particularly in long-term drought conditions.
7. Soil Health in the Context of Climate Change and Human Health: Discussing the evolving concept of soil health, which now includes considerations of water quality, climate change, and human health, and the challenges in quantifying soil health due to limited methods and knowledge about soil biodiversity.
8. Biological Complexity of Soil and Ecosystem Health: An exploration of the critical role of soil biology in nutrient cycling, soil structure, mineralization, and disease suppression, and its impact on plant growth, environmental quality, and ecosystem health.
9. Soil Phylogenomics and Agronomic Practices: Focusing on the knowledge gap in how agronomic practices and environmental processes alter soil microbial communities and how soil phylogenomics can be used to understand and enhance nutrient cycling and sustainable agriculture.
10. Innovative Bioremediation and Soil Microbial Dynamics: Investigating the dynamics of soil microbial populations in the context of bioremediation and how understanding these dynamics can contribute to soil health and sustainable agricultural practices.
Summaries, Data Governance and Analysis
Another great time saver is using LLMs to summarize large, complex documents. LLMs can also clean data and help standardize it. Sometimes, it takes a little coaching to get it to perform, which is troubling, but you can actually talk it into helping with a task when it balks.
Text to Image and Video
One of the most frustrating yet rewarding uses of Generative AI is prompting for images and video. Whether you use Adobe Firefly, Playground.AI, DALL-E 3, Stability.AI, Runway or Midjourney, one of the best ways to speed up the learning curve is to join Discord and find a channel for your preferred tool. A great way to test these tools is to use the same prompt on different platforms to see what fits your needs.
Wrap Up
Last thoughts on where we are headed as an industry. If we decide correctly, these tools promise to give us more time for the most fulfilling part of what we do. Deeper thinking, better strategy, more thoughtful expression, and, hopefully, more time to spend outside the virtual. Time to pursue what is truly good and beautiful.
This new reality is a future we need to take on together.