The Reality Check: AI's Disappointing Aspects in 2025
As we approach the midpoint of the 2020s, yes, music-lovers, it has been more than 30 years since Hootie and the Blowfish released “Cracked Rear View”, it's time for a candid discussion about artificial intelligence. While AI has undoubtedly made significant strides, it's crucial to address the disappointing aspects that have emerged. As a tech enthusiast living in the growing tech hub of Boise, Idaho, I've had a front-row seat to both the promises and letdowns of AI. Let's dive into the reality of AI in 2025.
Hype vs. Reality: The Expectation Gap
Remember all those promises of AI solving world hunger and curing cancer by 2025? Yeah, about that... The tech industry's tendency to overhype has led to unrealistic expectations. While AI has improved many aspects of our lives, it hasn't delivered the utopian future some predicted. It's a bit like moving to Boise for the perfect work-life balance – it's great, but it's not paradise.
The Bias Problem Persists
Despite years of awareness and efforts to address it, AI bias remains a significant issue. From facial recognition systems struggling with diverse faces to hiring algorithms favoring certain demographics, we're still grappling with how to make AI truly fair and inclusive. It's a reminder that AI, like humans, can inherit and amplify societal biases.
The Explainability Crisis
As AI systems become more complex, understanding how they reach decisions has become increasingly challenging. This lack of explainability is particularly concerning in critical areas like healthcare and finance. Imagine if The Wilder, my favorite local restaurant, couldn't explain what's in their dishes – you'd be hesitant to eat there, right? The same principle applies to AI we're supposed to trust with important decisions.
Data Dilemmas and Synthetic Solutions
High-quality data, the lifeblood of AI, is becoming scarcer. Companies are turning to AI-generated datasets, which introduce new biases and accuracy issues. It's like trying to teach someone about Idaho's outdoor beauty using only artificially created images – you'd miss the real essence of our breathtaking landscapes.
From Concept to Flop: AI Project Failures
Many AI projects, especially in the realm of generative AI, have failed to move beyond the proof-of-concept stage. Poor planning, unclear objectives, and lack of cross-team collaboration have led to disappointing outcomes. It's reminiscent of how Boise has grown rapidly without always having the infrastructure to support it – growth without proper planning leads to problems.
Where's the Wow Factor?
At recent tech events, AI applications have been described as "mostly meh." The initial excitement of AI has given way to a sense of stagnation in truly groundbreaking advancements. It's like visiting the same hiking trail in Idaho over and over – beautiful, but you start yearning for new discoveries.
The Road Ahead
The rapid advancement of AI has outpaced our ability to address its ethical implications. From the alignment problem to concerns about AI control, we're navigating uncharted ethical territory. It's crucial that we approach AI development with the same care and consideration we give to preserving Idaho's natural beauty for future generations.
Balancing Optimism with Realism
While it's easy to focus on the disappointments, it's important to maintain perspective. AI, like any transformative technology, is going through growing pains. Just as Boise has faced challenges with its rapid growth but remains a fantastic place to live, AI continues to hold immense potential despite its current shortcomings.
As we move forward, let's approach AI with a balanced view – optimistic about its potential but realistic about its limitations. By addressing these disappointing aspects head-on, we can work towards an AI future that truly benefits humanity, much like how thoughtful development can preserve the best aspects of a growing city like Boise.
What are your thoughts on AI's progress? Have you experienced any of these disappointments firsthand? Share your experiences in the comments below.
Probabilities, Predictions, and Updates: Lessons from the 2024 Presidential Race
In predictive modeling, probabilities offer an estimate of how likely an event is to occur, but they do not equate to certainty. This distinction is important in politics, as shown by the 2024 U.S. presidential race. Prior to Joe Biden’s unexpected decision to drop out, his probability of securing the Democratic nomination was high, based on polling data and historical trends. However, his exit forced analysts to update their predictions, causing the probabilities of other candidates like Kamala Harris and Gavin Newsom to rise. This illustrates how predictions must be updated in response to new information, reflecting the evolving nature of political forecasting. Understanding that probabilities reflect likelihood, not guarantees, helps manage expectations and allows for better decision-making in uncertain environments. In politics—and beyond—predictions are tools, not certainties, and must be regularly revisited as conditions change.
In the world of predictions, probabilities guide decision-making in finance, sports, politics, and everyday life. But as events unfold, we often need to update our beliefs based on new information. One of the most vivid examples of this process is in political forecasting, where the probability of a candidate winning an election shifts in response to changes in the political landscape. Let’s explore how probabilities work, when predictions should be updated, and how Joe Biden’s unexpected decision to drop out of the 2024 presidential race shifted the entire field.
How Probabilities Inform Predictions
Probabilities are estimates of how likely an event is to occur. In politics, these are often calculated by analyzing polling data, historical voting trends, demographic patterns, and economic conditions. In the lead-up to an election, experts assign probabilities to candidates based on current information. For example, if a candidate is polling strongly across key states, their probability of winning is higher than that of a candidate lagging in the same polls.
However, probabilities are not guarantees. Just because a prediction gives one candidate a 70% chance of winning doesn’t mean that outcome is certain. In fact, the remaining 30% probability still represents a real chance for the less-favored outcome to occur. This is where understanding the difference between probability and certainty becomes crucial.
The Balance Between Probability and Certainty
It’s easy to fall into the trap of equating high probability with certainty. For example, if a model predicts that a candidate has a 90% chance of winning, some people might assume that victory is a foregone conclusion. However, probabilities describe likelihood, not inevitability. Even a 90% chance of winning means there is still a 10% chance the outcome could be different—small but not negligible.
In politics, these small probabilities can have enormous implications. The 2016 U.S. presidential election is a classic case where many models showed a high probability of Hillary Clinton winning, but the less likely outcome—a Donald Trump victory—materialized. This demonstrates that no matter how confident a prediction seems, uncertainty is always present.
The challenge, then, is to reconcile probability with the human desire for certainty. In decision-making, it’s essential to recognize that high probabilities increase confidence but don’t guarantee results. By acknowledging that uncertainty is an inherent part of forecasting, we can make more balanced decisions, understanding that rare events do happen.
Prediction Updates: The Impact of New Information
Predictions must be dynamic because no model or probability can account for everything. This is where the concept of Bayesian updating comes in, a process in which predictions are revised when new data is available. Bayesian updating hinges on this formula: starting with an initial belief (called a prior), you adjust it based on how likely the new data would be if that belief were true.
Let’s take the 2024 presidential race as an example. In the months leading up to the primaries, Joe Biden’s re-election campaign seemed certain. Polling data consistently showed him as the frontrunner, and many analysts gave him a high probability of securing the Democratic nomination.
However, when Biden unexpectedly announced his decision to drop out, that probability changed instantly. The prediction of him winning the nomination dropped to zero, and analysts had to reassign probabilities to other Democratic candidates.
How Joe Biden’s Exit Shifted Probabilities
Biden’s exit triggered a significant shift in the political landscape, akin to removing the king from a chessboard. Other Democratic candidates who previously had slim chances—such as Kamala Harris, Gavin Newsom, or Pete Buttigieg—suddenly saw their probabilities skyrocket.
For example, if a candidate was polling at just 10% for the Democratic nomination while Biden was still in the race, their chances may have increased to 30-40% post-Biden. Political prediction markets, such as FiveThirtyEight or PredictIt, would have seen a rapid shift as analysts integrated new polling data reflecting Biden’s absence and re-evaluated the competitive dynamics of the race.
Republican probabilities also shifted, as Biden’s departure changed the equation for a general election. While many analysts had anticipated a Biden-Trump rematch, his absence opened the door for a more unpredictable scenario, increasing uncertainty in predictions for both sides.
When to Update Predictions: Key Considerations
1. Significant Events: A candidate dropping out, a major scandal, or a game-changing debate performance are all triggers for updating predictions. In Biden’s case, his departure was a critical event that forced an immediate recalibration of probabilities for all candidates.
2. New Polling Data: As fresh polling data comes in—especially after key events like primaries, debates, or campaign announcements—predictions need to be adjusted. A sudden surge in support for a candidate can significantly alter their chances of winning, as was the case when Barack Obama surged in 2008 following early primary wins.
3. Changing Fundamentals: Economic data, national crises, or shifts in voter sentiment can also necessitate updates. If the economy worsens or a major policy issue arises, the candidates’ probabilities will need to be reassessed based on how voters respond.
4. Time Sensitivity: The closer we get to Election Day, the more accurate our predictions should become, as the available data increases in volume and quality. Early in the race, predictions might be more speculative, but as the campaign progresses and more voters make up their minds, predictions tend to stabilize—although major events can still shift probabilities even at the last moment.
Learning from the 2024 Presidential Race
The 2024 presidential race provides a timely example of how dynamic predictions are and how quickly they can change. Biden’s decision to drop out forced political forecasters to re-evaluate the race in real-time, highlighting the importance of remaining flexible and updating predictions as new information comes in.
Probability vs. Certainty: What We Can Learn
The 2024 race demonstrates an important lesson: probabilities inform us about likely outcomes, but they do not guarantee results. Predictions are tools for understanding the future, not a crystal ball. The very nature of probability acknowledges the presence of uncertainty, which is why even a 90% chance of something happening still leaves a 10% possibility for a different result. Understanding the balance between probability and certainty helps us navigate complex situations with more caution and clarity.
Conclusion
The 2024 presidential race and Biden’s exit underscore a key truth about probability and predictions: they are not fixed points but evolving assessments of reality. By understanding how to update predictions and recognizing when to do so, we can make smarter, more informed decisions in an ever-changing world. Whether in politics, business, or life, accurate forecasting depends on staying adaptable, and the future depends on how well we learn to revise our beliefs in the face of new information.
About the Author: Curt Jones is the founding partner at Proklamate, a boutique, fractional analytics firm in Boise, Idaho. Proklamate specializes in business intelligence for small and medium-sized businesses.