When AI Becomes a False Prophet: A Cautionary Tale for Forecasters

Written by

When AI Becomes a False Prophet: A Cautionary Tale for Forecasters

With a nod to Taylor Swift and Travis Kelce, Superforecaster Ryan Adler discusses the gospel according to AI and why forecasters should always verify their sources.

Google’s AI Overview references an AI-generated video to support a false claim.

The promises of artificial intelligence have set up camp in media headlines over the past few years. ChatGPT has become a household name, billions are being spent just to power the equipment to run these programs and models, and the cutting-edge technology is front and center in ongoing tensions between the US and China. It hasn’t left any aspect of human activity untouched, including forecasting.

To be sure, the impacts already felt cannot be understated. We are looking at the front end in what I’m confident will be a seismic shift in society, with large swaths of labor markets around the globe being shaken to their core. That said, we aren’t there yet.

Here’s a recent example of how AI took itself out at the knees regarding a recent forecasting question on Good Judgment Open. In late April 2025, the time came to close a question regarding potential nuptials between Kansas City Chiefs star Travis Kelce and pop superstar Taylor Swift: “Before 19 April 2025, will Travis Kelce and Taylor Swift announce or acknowledge that they are engaged to be married?” (It’s not my favorite subject matter, but we try to maintain a diverse pool of questions.)

As a moderately rabid Chiefs fan myself, I was confident the answer was no, because that would have made headlines across media outlets. However, a key part of the job of running a forecasting platform is being in the habit of double and triple checking. So, I checked with Google. I entered “Are Travis Kelce and…” into the search field, which immediately autofilled to “are travis and taylor engaged?” (The first-name thing with pop culture stars annoys me to no end, but I digress.) To my surprise, Google’s AI preview popped up immediately.

“Yes, according to reports, Travis Kelce and Taylor Swift are engaged.”

“Trust, but verify”

Skeptical, I looked at what the experimental generative AI response was using as a reference to return such a statement. That’s when things got fun.

The first link of the cited material was a YouTube video. Keep in mind that Google, the search engine I used to start my research, owns YouTube. The account that posted the video? DangerousAI. That alone raises more red flags than a May Day parade in Moscow circa 1974. The brief video, dated 24 February 2025, purported to show Travis Kelce announcing that Swift and he “got engaged last week.” However, as the video progressed, the absurdity of Kelce’s putative announcement became perfectly clear.

To sum up, Google’s AI system linked to search was fooled by an AI product posted on another Google platform to give a patently false response.

I don’t highlight this incident as a criticism of Google. However, it should serve as a warning. I’ve seen some GJ Open forecasters take AI responses as gospel. I’m here to tell you that in matters of facts vs fiction, AI is very capable of being a false prophet. This is not to say that AI isn’t an incredibly valuable tool. It certainly is! We are finding more and more uses for it at Good Judgment, but we put it through its paces long before we deem it reliable for a particular role. As the Russian proverb instructs, “Trust, but verify.” (No, President Reagan didn’t say it first.) When it comes to AI and everything else you see online, my suggestion is that you just verify.

Do you have what it takes to be a Superforecaster? Find out on GJ Open!

* Ryan Adler is a Superforecaster, GJ managing director, and leader of Good Judgment’s question team

Meet the winner of the “Right!” said FRED: Q3 2024 Challenge

Written by

Meet the winner of the Q3 2024 “Right!” said FRED Challenge

The winner of the Q3 2024 “Right!” said FRED Challenge, Julio Vieiro, is a retired geologist with two decades’ experience in the oil industry as well as an investor, photographer, and painter. Known on GJ Open as JAVL, in this interview he discusses his interest in forecasting, his diverse hobbies, and his tips for fellow forecasters. He lives in Argentina.

GJO: Could you please tell us about yourself and your background?

Hi. I was born and live in Argentina. Professionally, I’m a geologist, and I have also studied biology and engineering. I worked almost two decades in the oil industry, and I’m currently retired. Additionally, I’ve been a photographer and painter for many years. Nowadays I dedicate my ever-scarce time to personal investments in the stock market, artistic activities, and a two-year program to become a professional sommelier. I enjoy mountains, the outdoors, sports, reading, and barbecues with my partner, my two adult children, and close friends. I’m interested in diverse topics.

GJO: How did you first become interested in forecasting? What brought you to GJ Open specifically?

During my professional development, I was always involved in the forecasting of technical parameters such as hydrocarbon saturation, mineralized thickness, porosity, pressure, etc. In this context, I was often curious why many highly trained professionals had an unconscious tendency to “fall in love” with their own ideas or models, or to expect results mainly according to their desires or what they thought someone expected, wrongly allocating resources or expectations. Frequently, people were more accurate evaluating their peers’ projects than their own.

In my case, Daniel Kahneman’s Thinking Fast and Slow was very revealing. It allowed me to give shape to the intuition that humans have evolved not to develop accurate objective explanations of reality but to generate quick, plausible explanations in order to effectively operate in our original environment, which is not the current one. The way I see it, many times our minds tend to take shortcuts that seem reasonable and/or comforting instead of doing the hard work of looking for evidence. This characteristic would not always help us understand complex situations, especially if we are emotionally involved.

Regarding GJ Open, during the pandemic I read Philip Tetlock’s Superforecasting and found it captivating. I decided to join GJ Open as a fun challenge to test in a non-subjective way my abilities to predict results of complex situations.

GJO: You had some tough competition in the Q3 2024 “Right!” said FRED Challenge. In your opinion, what helped you top the leaderboard?

I’m sure there’s always a dose of luck, some randomness. On the other hand, I try to pay attention to economic data. There is a great availability of information and projections, and it’s difficult to weigh which variables are really relevant and which have little or no impact. I also try to understand the probabilities of future events that could change the observed trends.

GJO: What types of forecasting questions do you enjoy the most? What topics would you like to see more of on the platform in 2025?

I’m more interested in some topics, such as economics, technology, space, sports. I tend to focus on what I enjoy most and what I’m interested in learning about. Evaluating the possible results helps understanding and stimulates the search for information. I prefer questions with data available, not just opinions. I also prefer questions where possible outcomes are expressed in ranges with multiple probabilities rather than yes/no or single probability answers.

I would like to have more questions related to the Latin American reality. In particular, my country is currently going through a complex and interesting economic and political process with still uncertain results.

GJO: What tips could you offer beginner forecasters on GJ Open?

I’m not sure what the best advice is, and I suppose there are many things that could be said. In any case, what I try to do is understand the question very well, not give an opinion on things that I am totally unaware of, look for reliable information and sources of data with frequent updates, and discriminate the relevant factors among the multiple ones that could influence a result.

What I find most important, however, is to find the point at which doubt or uncertainty makes us start to feel uncomfortable with the probabilities we assign. In my experience, people tend to feel too confident or secure. In many of these cases, something important could be missing. On the contrary, I often tend to overextend the alternatives to ensure I don’t err. I try, then, to force myself to rethink the problem in a different way or, more thoroughly, look for more information, or review what I have, to adjust the answer until I feel unsure. It’s not usually easy. Logically, at some point there are cases that are mathematically defined or where the probabilities seem to me less than 1/1000. Another exercise that I find useful is to assign a higher likelihood to those outcomes that don’t match the one that I prefer.

GJO: Is there anything else you’d like to add?

It seems to me that the exercise of making forecasts accustoms us to confronting our ideas with reality and trains us to make decisions with probabilistic logic, which I consider very important for both professional and personal life. It is also a very fun activity, a game where one competes against others and against oneself. Many thanks to Good Judgment for the enjoyment.

See the latest forecasting challenges on GJ Open and try your hand at forecasting!

Interview with the winner of the “Right!” said FRED Challenge

Written by

Interview with the winner of the “Right!” said FRED Challenge

In this interview, we sit down with the winner of the “Right!” said FRED Challenge for Q2 2024, Sigitas Keras. Known on GJ Open as sigis, Sigitas is an experienced quant and trader who decided to explore the world of forecasting after an impressive 25-year career in finance. With a PhD in mathematics and a natural curiosity about the world, he shares insights into the unique challenge he has taken on to forecast every question on GJO in 2024 and the strategies that helped him excel on the platform. Originally from Lithuania, Sigitas currently lives in Canada.

GJO: What is your background, and how did you first become interested in forecasting?

I was born in Lithuania, have a PhD in maths, but, as many others with a similar background, ended up in finance industry. After almost 25 years as a quant and a trader, I recently retired, which freed up a lot of time for other things. I tried forecasting on GJO for the first time a couple years ago. It seemed like an interesting challenge where I could combine analytical skills and general curiosity about the world.

GJO: How did you learn about GJ Open? How would you describe your experience on the platform so far?

I read Tetlock’s book Superforecasting, so likely that was an initial prompt, but to be honest I don’t remember full details anymore. Rightly or wrongly, I am one of the few forecasters who decided to forecast every question in 2024. It was very enjoyable, and I feel I learnt a lot both about forecasting and about various topics, but I have to admit this is getting too difficult to maintain. I don’t think I’ll continue doing all questions next year, and most likely will just focus on a few challenges, but I still like to maintain a good mixture of various topics.

GJO: What was your approach to the “Right!” said FRED Challenge? What do you think helped you come out on top?

I like questions that have good supporting data. In that sense, the FRED challenge is perfect for me. Whenever there is good data available, I try to use some mathematical model. Having a background in finance industry helps a bit with that, although I don’t think I use anything that requires more than FRED and other publicly available data and a Google spreadsheet. I also try to update my forecasts regularly, typically once a week. I think consistency is another important component of successful forecasting.

GJO: What topics would you consider of particular interest to forecast for 2025 and beyond?

I tend to forecast better when there is good data available for analysis. On the other hand, geopolitical questions are often much more challenging, so perhaps I will focus on improving there. My goal is to improve my score in the Superforecasting Workshops challenge!

GJO: Is there anything you would like to add that would be of particular interest to other forecasters on GJ Open?

I feel I am still very new to forecasting and to the community. One thing I hope is to learn more about other forecasters, their backgrounds, their approaches to forecasting. And if anyone has any questions for me, feel free to reach out.

See the latest forecasting challenges on GJ Open and try your hand at forecasting!