If you have spent the last few weeks testing out ChatGPT, the AI chatbot that has turned everyone on social media into a futurist, you might be impressed — as I have been — by its utter humility.

There’s barely any hint from the chatbot itself that many are touting its arrival as an “iPhone moment,” a technological revolution that many believe will disrupt multiple industries, impact millions of jobs, and fundamentally change the way we go about our work and life.

On futures and foresight, the focus of my own queries, ChatGPT readily admits that it’s not a crystal ball (a good start). As a language model, “it does not have the ability to predict future events or to understand the long-term consequences of actions.” Or so it tells me. It will even qualify any future-oriented questions with a standard disclaimer, “It’s hard to predict the future of [X] as it will depend on many factors…”

ChatGPT also confesses to other things it can’t do. It can’t understand what it says — neither the meaning nor concepts of its utterances. It can’t fact-check — at least not yet. By its own admission, it “does not have the ability to reason, plan or solve problems in the same way that a human can.”

All it does, after being trained on 570 GB of text, is generate conversational responses by predicting the next word in a sequence based on words that came before it. Somehow, that is enough to inspire people with awe. Over the last few weeks, we’ve witnessed impressive demonstrations of ChatGPT at work. We’ve seen it write code, do math homework, produce college essays and screenplays, answer complex questions on almost any topic, as well as generate blog posts and ad copy — things that fairly educated humans have been tasked to do, up till now.

For futurists, this technological breakthrough is raising the same troubling question that’s on everyone’s mind: Can AI chatbots do my job? More specifically, is ChatGPT capable of performing my job as a strategic foresight practitioner?

After taking ChatGPT on a test run, the short answer I would offer is no (at least not yet), but it can be helpful.

Image by Mark Frauenfelder, using Midjourney

First, its key limitations:

1. ChatGPT can’t easily generate signals of change. Because it was trained on data up till 2021, doesn’t search the internet in real time, and doesn’t have access to news sources, it can’t easily give you specific instances of how things could be remarkably different from today. It can speak generally about the rise of urban farming but not point you to the existence of a 26-story pig farm in China so that you understand its implications for automation, food production, and disease control.

2. ChatGPT tends to consider drivers in isolation. It can talk about what AI, aging population, or widening inequality might mean in the future but not analyze how various drivers might interact with one another. This might be because most industries tend to talk about trends in short and medium terms and that’s what ChatGPT is trained on as well. You would have to push it to integrate greater complexity.

That said, large language models in the form of AI chatbots can still be useful. When I asked for “a list of ten good questions that futurists might ask ChatGPT,” this is what the chatbot returned:

It is important to note that ChatGPT is a statistical model and is not able to predict the future with accuracy. Its output should be used as a starting point for further analysis and should not be taken as a definitive prediction of future events.

As you can see, they are all about the implications of a single driver or disruption. And if we choose one of the questions and pose it to ChatGPT, we get the following response.

Me: What are the potential social and economic impacts of widespread automation?

ChatGPT: The potential social and economic impacts of widespread automation are complex and multifaceted, and will depend on the specific technologies and industries that are automated. Some potential impacts of automation include:

Overall, the impact of automation on society and the economy will depend on how it is implemented and how the benefits of increased productivity and efficiency are distributed. It will also be important to consider policies and strategies to address the potential negative impacts of automation, such as job displacement and inequality.

For people who have considered this issue already, ChatGPT’s answer will seem fairly standard and predictable. It’s what I would call the consensus view, which in itself, is not unhelpful.

So how might foresight practitioners use ChatGPT in their work going forward? I would treat it first and foremost as a text generation engine and explore its utility in the following ways:

A good rule of thumb is to treat ChatGPT as the starting point, not the final word. It can be helpful when investigating a domain you’re not familiar with. But you will still need to bring a critical eye to fully develop the plausible futures that can guide us to make better strategic decisions today.

#foresightmethods #futuresthinking #foresight

Institute for the Future (IFTF) is the world’s leading futures organization. Its training program, IFTF Foresight Essentials, is a comprehensive portfolio of strategic foresight training tools based upon over 50 years of IFTF methodologies. IFTF Foresight Essentials cultivates a foresight mindset and skillset that enable individuals and organizations to foresee future forces, identify emerging imperatives, and develop world-ready strategies. To learn more about how IFTF Foresight Essentials is uniquely customizable for businesses, government agencies, and social impact organizations, visit iftf.org/foresightessentials or subscribe to the IFTF Foresight Essentials newsletter.