Stop 1: “The Trailhead”

Getting a Lay of the Land

We Used ChatGPT To Help Us With Trend Investigation. Here’s What We Found.

In a previous update, we introduced the exciting journey Sundberg-Ferar is currently on, undertaking a mock project (based on the National Parks Michigan Mobility Challenge) to explore the capabilities of cutting-edge artificial intelligence tools so we can uncover the most effective ways to apply them.

Now, as we dive into the project, our first task has been to immerse ourselves in the wealth of information on national park mobility that is already publicly available to guide and inspire innovative thinking in the phases to come. This means uncovering sources of data, reports, and macro trends that may impact design, technology, and the stakeholders of our eventual solution.

Given our goals in this phase, we’re heavily utilizing ChatGPT-4, comparing it to its predecessor 3.5, which we were already familiar with (and which is currently the “free” version available for all to use). At the same time, our team is also experimenting with other relevant tools – such as Perplexity – but to start we’ll be mainly focusing our attention on the benefits and shortcomings discovered through our deep dive with ChatGPT. Stay tuned for more updates as we continue to explore diverse AI tools throughout our project journey!

The advancements between versions 3.5 and 4 are clear, as would be expected with a free versus paid service. Some of the first improvements our team experienced were enhancements in being able to understand situational context, the ability to produce source citations for information and data that appears in responses generated by the tool, and the ability to upload and analyze an array of file types.

1. Document Examination Assisted in Refreshing Insights

Our team curates and maintains a trend library of the most up-to-date macro trends we’ve identified across a multitude of industries and topics. Watching trends has long played an important role in our design and innovation processes at SF – putting on a future-focused lens allows us to help our clients understand who their customers might be into the future, while at the same time anticipating their needs in the context of new social, technological, environmental, and geopolitical realities. However, across the hundreds of trends we’ve gathered, it can be a daunting task to keep each up-to-date with the most recent industry data.

Using ChatGPT-4 to analyze our library has been a revelation. By uploading a PDF of trends in our library that we felt related to national park mobility, the tool was not only able to comprehend the document’s contents, but our team was also able to prompt it to provide recent and relevant data sources to update and refresh individual trends.

For example, we had a trend related to the aging population in the United States, supported by data from 2021. While not extremely out-of-date, we felt sure there would be more recent data available on national aging. Sure enough, ChatGPT was able to immediately point us toward Census data from 2022 to refresh our statistics. This feature significantly streamlines our information-gathering process – helping us understand if the data we are seeking might exist, and where we can go to access it.

2. Generative Thinking Pointed Us in the Right Direction

While we curate and maintain a library of macro trends, we also always curate custom trends related to unique project goals and stakeholders. In prompting ChatGPT-4 to assist with this effort, we were impressed with the quantity of topics the tool was able to identify for consideration. Its responses were organized logically – providing overarching themes (such as “Smart Technology and IoT”) as well as associated impacts to consider (“IoT devices for monitoring and managing traffic flows, parking, and visitor experiences”).

It quickly became apparent though that these outputs would mainly serve as a starting place – topical inspiration requiring more thorough exploration by our team to understand how novel and pertinent they truly are to the project domain, as well as their lasting impact which is a key characteristic of a trend that’s up-and-coming, as opposed to a trend that is waning or is really just a fad.

We also found that it took additional prompting to get the tool to “think” more divergently, identifying trends that might have an indirect impact on our project. But when our prompts were explicit about the lens we wanted the tool to analyze through, we were again pleasantly surprised with the results. For example, our team often curates social and behavioral shifts, to understand how the actual consumers of tomorrow might differ from what our clients know about their consumers of today. By asking for things like “demographic” and “psychographic” trends, we were able to explore even more topics, (such as “Remote work and digital nomadism” leading to visitors who may be seeking longer stays in national parks, and amenities like workspaces nearby).

The takeaway here was that iterative collaboration between our human team and ChatGPT gave us an immediate starting place of a diverse set of topics to explore, ultimately making us more efficient. However, the balance between AI assistance and including deeper human intuition and understanding of the market and broader societies was critical to this task’s success.

3. Analyzing Datasets Helped Us Understand Usage Patterns

To better understand current national park mobility initiatives and challenges, our team downloaded publicly available datasets, such as one set on visitation numbers at parks throughout the year. This came in the form of an Excel document, with more than 4000 entries in a given year, tracking the park name, month of the year, and visits counted in that month.

By uploading this document into ChatGPT-4, we were able to get immediate insights from the tool, including the average number of visits parks receive, the standard deviation of visits along with insights that the large deviation indicated a significant variation in visit numbers among different parks, as well as identification of the parks with the least and most visits. The tool even produced a helpful chart to visualize how visits vary over the year, with analysis of peak and off-peak months.

The one stumbling block ChatGPT ran into was with a prompt to analyze parks specifically located in the state of Michigan. Its original response referred to several parks clearly NOT located in the state (e.g., Yosemite National Park, San Antonio Missions National Historical Park). When prompted about the error, the tool was able to identify that it made the mistake, and correct it, producing results that did align with expectations when checked by a human against the original Excel file. However, this reiterates the point that human logic, reasoning, and intuition are still essential when utilizing AI tools to evaluate the validity of the outputs.

Looking Ahead: The Journey Continues

As we wrap up this task, the journey is far from over. We’ve scratched the surface, but the vast landscape of AI tools and their applications lies ahead. We hope you’ll join us on this expedition and together we’ll navigate the evolving landscape of AI, unlocking its potential for innovation and efficiency. Stay tuned for the next chapter in our AI adventure!

We invite you to join us on this adventure by subscribing to our newsletter and following us on Linkedin where we’re posting regular updates on what we’re learning, what shocks us disappoints us, delights us, and what will change how we approach innovation. 

Discover more articles on Innovation Strategy and Prototyping