My brief chat with ChatGPT 4 on the pitfalls of year-end predictions

December brings the barrage of an inescapable seasonal ritual.

Of course, I’m talking about the tendency for people—many perhaps too woozy on eggnog — to try to sum up the previous year and/or predict the year ahead.

Lately, I’ve obsessed about how we’re drowning in commentary and opinions on every platform — including TikTok, now the hot news source — even as we’re starved for rigorous original reporting among the rampant closure of local newsrooms.

So, I turned to ChatGPT for solace.

OpenAI, the company behind ChatGPT, provides a “ChatGPT Plus” subscription service with access to ChatGPT 4, a significantly better AI than the public version. I subscribed to experiment and have been surprised about how my prompt threads have felt more like conversations with an emotionally vulnerable entity than sterile bot chats. (Yes, OpenAI just navigated a radical leadership crisis. Listen to this episode of the New York Times podcast “Hard Fork” for an interview with CEO Sam Altman just before he was fired and then quickly rehired.)

This brief convo about predictions was no different. The AI still loves to churn out listicles, but its defense of critical thinking nearly brought a tear to my eye.

And ChatGPT may have sounded more human than ever to me thanks to this tiny, telling detail: It felt compelled to end our conversation with reassuring exclamation marks at the ends of sentences. If that’s not the sign of an insecure sentient being, I don’t know what is.

The Q&A that follows was edited only for style. (Mainly I was offended by its capitalization of “Black Swan.”)

My prompt: What do people — including journalists, commentators, and online influencers — tend to get wrong when making predictions at the end of the year about the year ahead?

ChatGPT 4: When making predictions about the upcoming year, people, including journalists, commentators, and online influencers, often make a number of common mistakes:

1. Overconfidence in predictions: There’s a tendency to be overly confident in one’s ability to predict the future. This overconfidence can lead to bold, certain statements about what will happen, without acknowledging the inherent uncertainty of future events.
2. Recency bias: People often focus on recent events or trends and extrapolate them into the future without considering that circumstances can change rapidly. This bias can lead to predictions that are too heavily influenced by what’s currently happening, ignoring historical trends or potential disruptions.
3. Confirmation bias: Individuals may selectively look for information that confirms their existing beliefs or desires, leading to skewed predictions. This can result in ignoring contrary evidence or alternative scenarios that might be more likely.
4. Underestimating complexity and interconnectedness: The world is complex and events are often interconnected in ways that are not immediately obvious. Simplistic predictions fail to account for this complexity and the myriad factors that can influence outcomes.
5. Anchoring to conventional wisdom: There can be a tendency to stick close to the consensus or conventional wisdom, leading to a lack of original thought or insight in predictions. This can result in echoing the same ideas and not considering outlier events or unconventional possibilities.
6. Neglecting “black swan” events: Unpredictable, high-impact events (known as black swan events) are often overlooked in predictions. These events, by their nature, are hard to predict but can have significant consequences.
7. Focusing on specifics over trends: Sometimes, predictions get too caught up in specific details rather than looking at broader trends. This can lead to missing the larger picture or misinterpreting the significance of certain events.
8. Influence of personal or political biases: Personal or political biases can heavily influence predictions, leading to a skewed perspective that may not accurately reflect reality.
9. Lack of expertise in relevant fields: Commentators may make predictions in areas where they lack deep expertise, leading to oversimplified or inaccurate forecasts.
10. Failure to update beliefs: As new information becomes available, it’s important to adjust predictions accordingly. However, many fail to update their beliefs and predictions in light of new evidence.

These common pitfalls highlight the challenges in making accurate predictions about the future and underscore the importance of critical thinking, open-mindedness, and humility in forecasting.

My prompt: What is one of the more egregious examples of somebody falling prey to one of these common pitfalls?

ChatGPT 4: One of the more egregious examples of someone falling prey to these common pitfalls in prediction-making is the widespread belief in the late 1990s and early 2000s that the housing market in the United States would continue to rise indefinitely. This belief was a classic case of recency bias, where recent trends in rising home prices led many experts, commentators, and financial professionals to assume that this trend would continue without a significant downturn.

Key mistakes made in this instance included:

1. Overconfidence in predictions: Many financial experts and commentators were overly confident that the housing market was invulnerable to a major crash, despite historical evidence suggesting that no market rises indefinitely.
2. Recency bias: Predictions were heavily influenced by the recent trend of increasing house prices, with many ignoring historical cycles of booms and busts in real estate.
3. Underestimating complexity and interconnectedness: Few predicted how deeply interconnected the housing market was with financial markets, particularly through mortgage-backed securities and other financial instruments. The complexity of these products and their widespread impact on the global economy were largely underestimated.
4. Neglect of black swan events: The eventual crash in 2007-2008, often considered a black swan event, was largely unforeseen by mainstream predictors. The severity and global impact of the housing market crash caught many by surprise.

This instance illustrates how a combination of overconfidence, recency bias, and a lack of understanding of market complexities can lead to significantly flawed predictions with far-reaching consequences. The 2008 financial crisis, which was partly a result of the housing market crash, had a profound impact on the global economy and serves as a cautionary tale about the dangers of erroneous forecasting.

My prompt: Taking advantage of the most recent data you have available, what do you think would be an example of predictions now being made about the near-term future?

ChatGPT 4: Based on the most recent data and trends up to April 2023, here are a few examples of predictions being made about the near-term future:

1. Technological advancements: There is a widespread prediction that advancements in artificial intelligence, machine learning, and automation will continue to accelerate. This includes developments in areas like natural language processing, autonomous vehicles, and personalized medicine. However, there’s also concern about the societal and ethical implications of these technologies, including job displacement and privacy issues.
2. Climate change and sustainability: Predictions around climate change remain a significant focus, with many experts forecasting increased impacts from global warming, such as more extreme weather events, rising sea levels, and biodiversity loss. Concurrently, there’s a growing emphasis on the transition to renewable energy sources and sustainable practices to mitigate these effects.
3. Economic trends: Given the global economic fluctuations experienced in recent years, including the impact of the COVID-19 pandemic, there are varied predictions about economic recovery and growth. Some forecast a steady recovery, while others predict potential economic downturns or recessions in certain regions, influenced by factors like inflation, supply chain disruptions, and geopolitical tensions.
4. Health care and biotechnology: The rapid development of vaccines and treatments during the COVID-19 pandemic has led to predictions about similar fast-paced advancements in health care and biotechnology. This includes the development of new vaccines, gene therapies, and digital health technologies.
5. Geopolitical shifts: Ongoing geopolitical tensions in various parts of the world, including conflicts and trade disputes, have led to predictions about changes in global alliances, economic policies, and power dynamics. This includes the evolving roles of major powers like the United States, China, and Russia on the global stage.
6. Remote work and the future of workplaces: The shift toward remote work and flexible work arrangements, accelerated by the pandemic, is expected to continue, with predictions about long-term changes in how and where people work. This includes the potential for a more distributed workforce and changes in urban planning and real estate.
7. Cybersecurity concerns: With the increasing digitization of many aspects of life, there are predictions of heightened cybersecurity threats, including more sophisticated cyber-attacks and data breaches.

It’s important to note that while these predictions are based on current trends and data, they are subject to change due to unforeseen events or shifts in global dynamics. As with any prediction, there’s a degree of uncertainty, and future developments could alter these forecasts significantly.

My prompt: To be clear, are you saying that the concern being expressed about each of these seven items may be overblown — or at least exaggerated — because the predictions made are too bleak? And is it just a coincidence that the No. 1 item involves artificial intelligence, or do you think your own analysis in some way may be biased because you represent a significant technological advancement?

ChatGPT 4: To clarify, the predictions and concerns listed about these seven areas are not necessarily overblown or exaggerated; rather, they reflect current trends and common themes in public discourse as of my last update in April 2023. The intention was to provide an overview of areas where significant attention and forecasting are being focused, rather than to suggest that these concerns are necessarily too bleak. Each area has its own set of complexities, and the level of concern or optimism can vary widely among experts and commentators.

Regarding the listing of artificial intelligence (AI) as the first item, this placement doesn’t necessarily indicate bias or suggest that it’s the most important or concerning area. The order was more about structuring a diverse list that covers various aspects of global interest, including technology, environment, economy, and social changes. AI is a significant area of development and public interest, especially considering its rapid advancements and potential impact across multiple sectors. Hence, it often appears prominently in discussions about future trends.

As an AI developed by OpenAI, I am programmed to provide balanced, informative, and unbiased responses to the best of my ability. My responses are based on the data and training I have received, which includes a wide range of sources and perspectives. While AI and machine learning models like myself are designed to be as objective as possible, it’s important to acknowledge that no source of information, AI or human, can be entirely free from all forms of bias. This is why critical thinking and considering multiple sources of information are always recommended when evaluating predictions and trends.

My prompt: That last paragraph was elegantly phrased and deployed. Thank you.

ChatGPT 4: You’re very welcome! I’m glad you found the response helpful. If you have any more questions or need further information, feel free to ask. I’m here to help!