Winging It in the AI Apocalypse
By Matt Jackson
Generative AI emerged with the suddenness of a large earthquake and with implications that shook our foundation. AI tools surfaced as simultaneously powerful and destabilizing, with inherent ethical and existential questions. I work in Technology and Innovation at a small liberal arts college, and am among the staff tasked with learning and, soon, training others in the ever-changing field that barreled onto our campus.
I quickly accepted that I cannot be the expert with all the answers. What I can do is equip myself and others with tools and foundational knowledge. Learning AI means relearning AI as it continues to evolve in the weeks, months, and, most likely, years to come. But before learning, we must deal with an all-too-frequent first step: how do you begin?
Negotiating Mnemonics
Attempts to develop tools to springboard myself and others into the world of AI started off with lots of floundering. How do I encourage use that is effective, while still conveying the flaws and limitations of today’s technology? Through lots of pain and denial, I realized I needed something small, transferable across forms of communication, and most importantly memorable.
Enter P.I.G.E.O.N. It’s important to note that I’ve never been a huge fan of mnemonics, but I can’t deny their efficacy in the right situation. But before we get too far, let’s address the elephant bird in the room—why pigeon?
For about a year and a half I’ve joked that AI tools are a great way to completely lose our own voices, and break down any form of communication we have. So, if we don’t get this whole AI thing right, I suppose we can always go back to carrier pigeons. It was in this off-hand joke that I found a word both workable and memorable. Even more important, I wanted to capture specific themes in this mnemonic.
Prompt Engineering offers limited returns. While strategies exist, they mitigate limitations and errors rather than perfect AI tools. Let's focus on utilizing these underlying strategies without implying they're revolutionary or require extensive training to use.
Proceed with caution. There are countless headlines pointing out terrible uses of these tools. These misuses often stem from ignoring major limitations or using the tools in unintended ways. Every user needs to constantly evaluate how and when they use AI tools.
Adaptation is paramount. AI tools are still works in progress. As with any technology, change is the only certainty, and AI is changing at a breakneck pace.
Failure is always an option. People often fear failing with technology. These tools take practice, so let's reassure users that it's okay to keep experimenting.
P.I.G.E.O.N.
P - Prepare
Before you start engaging with an AI tool, set realistic expectations. There’s still plenty AI can’t do, so don’t feel the need to force out a finished product. Additionally, these tools will almost always do better at editing your own content, rather than generating new ideas on their own.
I - Identify
Define a role when prompting the AI. You may give your own role, or ask the AI to roleplay as a specific role. This helps build important context to the work ahead. Make sure to clearly identify objectives and key contexts for your work.
G - Guide
You’re going to get weird things back from the AI sometimes. Remember that these tools are predicting the next words in a sentence. Make adjustments as the AI responds, and provide additional context if it’s needed to get the tools working in the right direction.
E - Experiment
Always keep iterating and trying again. Keep the speed of these tools in mind. If things go wrong in an iteration you’ve likely only lost seconds, not minutes, of work.
O - Observe
Critically read and engage with the outputs. This is imperative when we discuss issues of hallucination and inherent biases within training sets. Along those lines, recognize and acknowledge what tasks the AI is doing for you. Is ChatGPT “just editing a blog post” for you, or have you lost your own voice along the way? That’s a question we each have to answer for ourselves.
N - Navigate
Adapt as these technologies change. Part of my role is to deliver presentations about AI, and the major companies developing generative AI have an uncanny knack for publishing major updates only days before each presentation. Stay plugged in to the news, and be prepared to change your strategies as technology shifts.
Nest-Gen Approach to Next-Gen Tech
The P.I.G.E.O.N. framework gives a starting point for using these tools responsibly and effectively. As we continue to explore their potential, it’s crucial to approach with both enthusiasm and caution.
Remember, the goal isn’t to be a prompt engineer or AI expert. Instead, we're aiming to develop a thoughtful approach to these powerful tools, ensuring they enhance, rather than replace our unique human perspectives and creativity.
Matt Jackson is an Application Analyst at Davidson College, passionate about the intersections of technology on teaching and learning, with a recent focus on the advent of generative AI.