S: In the United States, around three million golfers pay an average of $33 each to maintain a World Handicap Index through a United States Golf Association (USGA) certified golf club. This amounts to an annual revenue of roughly $100 million—a sizable sum that essentially powers the operations of Allied Amateur Golf Associations (AGAs). While this is an estimate, it is within a reasonable range and serves to underline a potentially worrying correlation: should the proprietary World Handicap System (WHS) “formula” cease to be the “industry standard or benchmark”, AGAs may face an uncertain future.
There’s a counterpoint that the golfer’s fees mainly relate to AGA memberships, however, empirical data gathered from polls suggest a different view: over 90% of golfers incur this cost mainly to acquire a WHS Index.
Nevertheless, this post isn’t meant to instigate a debate over membership fees—a topic that can be reserved for another time. Rather, this post aims to draw attention to a specific idea: the potential consequences of an Artificial Intelligence (AI) based formula bettering the WHS formula and transforming golf handicapping!
Let’s consider this from a historical perspective. Looking back 50 years (which I can), handicapping scores were first recorded on paper posting sheets, then in proprietary software applications and/or kiosks, followed by websites, and more recently on mobile devices. Now, it’s watches (actually, doesn’t have to be a watch – think in terms of any device that is better than the device that preceeds it). Beyond collecting scores (mainly hole-by-hole), watches have a multitude of features including GPS to provide shot distances. Soon, AI powered watches will be conversing with the golfer, advising on which club to select and whether to go for the green or play it safe. They’ll also remind golfers how to best approach a shot based on past performance and offer advice on their setup, contingent on the type of shot being played. Think of a virtual pro that is familiar with both the golfer and the course (Lessons for the Jetson’s!).
Golf watches already compile data and supply most every conceivable statistic a golfer can imagine. But in the not-too-distant future, they’ll affect an accurate AI handicap and precise AI handicaps for other players in a tee group. That way games will be played, not on golfers scoring potential, but based on the score history of the group from rounds played together. In this future, where does an outdated WHS handicapping formula fit in (especially considering its reliance on humans to rate courses every 10 years)? Answer: It doesn’t.
Then there is this consideration – tens of millions of dollars and tens of thousands of man hours have been invested in the USGA/WHS formula, which has resulted in a fair amount of intellectual property! Does AI care? No. Let’s face it, when AI is focused on the task, it will create something better in less than a day!
S: Define and explain the differences between a predictive golf handicap system and one that determines a golfer’s potential.
ChatGPT: Certainly, a predictive golf handicap system and one that determines a golfer’s potential fundamentally differ in their objectives and the aspects of a golfer’s performance they emphasize.
- Predictive Golf Handicap System: The primary goal of a predictive golf handicap system is to forecast how a golfer is likely to perform in a future round of golf based on their past performances. Such a system uses past data – typically, a golfer’s previous scores – to compute a handicap that aims to predict their next score. This system usually involves complex statistical calculations or even machine learning algorithms to create a model that can effectively predict future performance. It assumes that past performance is a reliable indicator of future results.
- Golf Handicap System Determining Golfer’s Potential: This kind of system, on the other hand, is designed to estimate a golfer’s potential or their ability to perform under optimal conditions. Rather than trying to predict a specific future score, it aims to quantify the golfer’s best or peak level of performance. Typically, it might consider a golfer’s best scores from recent rounds to compute the handicap. The underlying assumption of this system is that a golfer’s best performances offer the most accurate reflection of their potential.
The fundamental difference between these two systems lies in their purpose and the performance aspects they emphasize. While the predictive system is concerned with forecasting a golfer’s likely score in their next round based on a broader set of past performances, the potential-oriented system seeks to quantify a golfer’s peak capability by focusing more on their best performances.
S: Given a large data set that includes golfers and their score records, could AI develop a golf handicapping formula that is more accurate than the World Handicap System while not using any of the USGA’s intellectual property?
ChatGPT: Given a sufficiently large and diverse dataset of golfers and their score records, it’s theoretically possible for a well-implemented AI to develop a golf handicapping formula that could potentially outperform the World Handicap System (WHS), without directly using the USGA’s intellectual property.
An AI-based system could employ advanced machine learning algorithms and statistical methods to discover patterns and relationships in the data that may not be captured by the WHS. The success of this approach would largely depend on the quality and representativeness of the data used, the sophistication of the AI algorithms employed, and the expertise of the humans overseeing the process.
S: If an AI-driven handicap system is better at predicting a golfer’s next score than does a handicap system that determines potential (WHS), wouldn’t it be better to use for tournament play?
ChatGPT: Whether a predictive handicap system is better for tournament play than a potential-based system largely depends on the goals of the tournament and the specifics of the handicap systems in question.
In theory, a handicap system that’s better at predicting a golfer’s next score could lead to fairer matchups in tournament play, as it could provide a more accurate reflection of each golfer’s likely performance in the upcoming rounds. It could potentially level the playing field more effectively, as it adjusts for each player’s current skill level based on their most recent performances.
However, a potential-based system has its advantages too. It recognizes a golfer’s ability to perform under optimal conditions, which might be more appropriate for a tournament setting where golfers are expected to bring their best game. This could encourage players to strive for their best, knowing that their peak performance is what’s being measured.
Another factor to consider is how players and organizers perceive the fairness and understandability of the system. Even if an AI-driven system is technically more accurate at predicting scores, it could be seen as a “black box” if the players and organizers don’t understand how it works. This could lead to concerns about fairness and transparency, which are crucial for the acceptance of any handicap system in tournament play. So, while an AI-driven handicap system that’s better at predicting scores could potentially be useful for tournament play, it’s not a straightforward decision and would depend on various factors, including the goals of the tournament, the specific characteristics of the handicap systems, and the perceptions of players and organizers.
S: A large majority of golfers don’t care about the black box aspect of handicapping, they trust technology. Taking this into account, do you have more to add?
ChatGPT: Certainly, if the primary concern of golfers is trust in the technology and its ability to produce fair and meaningful handicaps, then an AI-based system could potentially be an appealing option. This perspective significantly diminishes the importance of transparency and interpretability, two aspects that can be challenging with AI.
If an AI-based system were built using high-quality, comprehensive data, it could offer a potentially more accurate and nuanced measure of golfers’ abilities and the likelihood of their future performance. Furthermore, given the adaptive nature of AI models, such a system could continuously improve its predictions based on the influx of new data, making it a potentially more dynamic and responsive system compared to traditional methods.
Also, an AI-based handicap system could incorporate a broader array of factors that influence a golfer’s performance, such as detailed course conditions, weather variables, and player-specific characteristics. This could result in more personalized handicaps that take into account a wider range of performance influencers than the WHS does.
In conclusion, given that the majority of golfers trust technology and are less concerned with the “black box” nature of AI, an AI-based handicap system could indeed present a promising avenue for potentially improving upon the current WHS, given the availability of adequate resources and high-quality, comprehensive data.
S: If an AI handicap system is better at predicting a next score shot than does a system that determines potential (like the WHS), couldn’t people use the predictive information to better determine who might win in a tournament that uses a handicap system based on potential?
ChatGPT: In theory, a predictive AI handicap system that is more accurate at forecasting a golfer’s next score might indeed provide valuable information when attempting to speculate on the outcome of a tournament. By comparing the predictive handicaps to those assigned by the potential-based World Handicap System (WHS), spectators, bettors, or even the players themselves could potentially gain insights that might not be evident from the WHS handicaps alone.
S: If a golf handicap system was charged to hit a certain degree of accuracy, wouldn’t the formula become less relevant, and the governance be more relevant? For example, we have an FDA that determines if a drug is acceptable before it’s made available to the public. Couldn’t there be the same oversight for handicapping?
ChatGPT: Indeed, if a golf handicap system were tasked with achieving a specific degree of accuracy, the focus might shift from the formula used to create the handicap to the governance of the system. The governance could involve the development of standards and regulations, the enforcement of these rules, and the evaluation of the system’s performance and accuracy. This perspective is similar to the role of the FDA, which oversees the safety and effectiveness of drugs, rather than developing the drugs themselves.
Under such a governance model, the role of a governing body could be to set the standards for what constitutes an acceptable level of accuracy, to establish guidelines for the development and operation of the handicap system, and to conduct or oversee evaluations of the system’s performance. It might also involve ensuring that the system is fair and does not disadvantage certain groups of golfers.
In the case of an AI-based handicap system, the governance could involve standards for data collection and processing, guidelines for AI model development, and rules to ensure the privacy and security of golfer data. The governing body could conduct or oversee evaluations of the AI model’s performance, perhaps by comparing its predictions to actual golfer scores.
This kind of governance approach could help to ensure that the handicap system is accurate, fair, and trustworthy, regardless of the specific formula or model used to calculate handicaps. However, it would also require considerable resources to establish and maintain such a governance system, including expertise in golf, statistics, and possibly AI and data privacy.
In conclusion, a governance model could potentially be a useful approach for ensuring the accuracy and fairness of a golf handicap system, whether it uses a traditional formula like the WHS or an AI model. However, implementing such a model would likely require significant resources and expertise.
S: The USGA tests golf balls and golf clubs so that they conform to certain limits. Couldn’t the USGA have a set of golfer and score data that it could provide to AI-based handicap systems and as long as they prove their accuracy they could be certified. Thoughts?
ChatGPT: Indeed, this could be a feasible and efficient approach for the USGA to take if they wanted to oversee AI-based handicapping systems. By providing a standard dataset to AI developers, the USGA could create a common benchmark against which all systems are tested. This would help ensure consistency in evaluations and make it easier to compare the accuracy of different systems.
To implement this, the USGA could assemble a comprehensive, anonymized dataset of golfer scores and associated data, such as course conditions, golfer statistics, etc. AI developers could then use this dataset to train and test their models. Once their systems have been developed, they would provide their handicap predictions for a set of golfers in the dataset. The USGA would then compare these predictions to the actual scores to assess the system’s accuracy.
This approach has several advantages. It would provide a fair and consistent way to evaluate different systems, and it would allow the USGA to control the quality and appropriateness of the data used for training and testing. It would also give the USGA a straightforward way to check a system’s accuracy without needing to understand the complexities of its underlying AI model.
S: If a large technology vendor got behind an AI handicap system would it help with acceptance?
ChatGPT: The backing of a reputable and influential technology vendor could indeed facilitate the acceptance of an AI handicap system in the golf community. This is due to several factors:
- Credibility and Trust: A known tech giant’s endorsement can lend credibility to the system and increase trust among users. People tend to trust brands they know and respect, and this can accelerate the adoption of a new technology.
- Resources: A large tech company has significant resources to invest in the development, refinement, and marketing of the AI handicap system. They can also afford to conduct extensive testing and validation to ensure that the system is accurate and reliable.
- Technical Expertise: A prominent technology company is likely to have considerable expertise in AI and data analytics, which can help in the creation of a sophisticated, effective handicap system.
- Marketing and Reach: A tech giant has a large platform and customer base, which can be leveraged to promote the AI handicap system and reach a wider audience.
- Integration with Existing Tech: A large tech vendor could potentially integrate the AI handicap system with other golf-related tech products and services, creating a comprehensive solution that adds value for users.
S: If the PGA of America were to get behind an AI handicap system would that help with acceptance?
ChatGPT: Yes, the endorsement of an esteemed organization like the PGA of America could certainly facilitate acceptance and adoption of an AI-based handicap system. The PGA of America is a respected authority in the golf world, and its support could lend significant credibility to the system.
Here are several reasons why their backing could be beneficial:
- Authority and Credibility: The PGA of America’s approval would provide the system with a level of authority and credibility. Their endorsement signals to golfers and other golfing organizations that the system has been evaluated and deemed trustworthy by a respected authority in the field.
- Widespread Exposure: The PGA of America has a broad platform that reaches a large audience of golf professionals and enthusiasts. Their support would ensure the system receives considerable attention and visibility.
- Implementation in Tournaments and Events: If the PGA of America were to implement the AI-based handicap system in its own tournaments and events, this could serve as a strong testament to its effectiveness and reliability, encouraging other organizations and individual golfers to adopt it as well.
- Education and Understanding: The PGA of America could play a vital role in educating golfers about the AI-based system, how it works, and its benefits. This could help overcome any resistance due to lack of understanding or mistrust of AI.
- Alignment with Rules and Regulations: An endorsement by the PGA of America would also suggest that the system aligns with golf’s rules and regulations, addressing any concerns about its legitimacy or fairness.
______________________________
S: AI’s ability to develop a more accurate (better?) handicapping formula and the USGA’s response will be one of the more fascinating threads within this blog! Will the USGA be able to contain AI, and if necessary, put it down? The intersection of governance, business, and AI will be fun to explore. Much more on this to come!
______________________________
S: One last request – Summarize AlphaZero’s mastery of chess.
ChatGPT: AlphaZero is a groundbreaking AI developed by DeepMind, a Google-owned artificial intelligence company. It rose to prominence in 2017 for its mastery of chess, along with two other complex games – shogi (Japanese chess) and Go.
AlphaZero’s unique approach is what sets it apart from previous game-playing AI. Instead of being pre-programmed with the best strategies or moves, it is designed to learn from scratch through a process called reinforcement learning. It starts by playing games against itself, making random moves at first. With each move, win, or loss, it adjusts its strategy, learning what works and what doesn’t.
After playing millions of games against itself and learning from each one, AlphaZero mastered the game of chess. Remarkably, it reached superhuman performance in chess in a remarkably short time – just a few hours of self-play. It then went on to defeat the reigning computer chess champion, Stockfish, winning 28 games, drawing 72, and losing none out of 100 games.
AlphaZero’s mastery of chess is seen as a significant achievement in the field of artificial intelligence. Its learning approach mirrors the way humans learn, suggesting that AI can adapt to new situations and solve problems without relying on pre-existing knowledge.
S: Compare your abilities to those of AlphaZero. Is this something to think about or is it better to let someone else think about it? 🤔
More on AlphaZero – https://en.wikipedia.org/wiki/AlphaZero
A worthwhile read – The Age of AI and Our Human Future
