Back to all posts

AI Revolutionizing Casino Slot Management

2025-05-23Buddy Frank, CDC Gaming Email, LinkedIn, and moreShare this article8 minutes read
Artificial Intelligence
Casino Management
ChatGPT

AI Robot Concept

The Initial Test: ChatGPT as a "Digital Freshman"

Just about two years ago, in September 2023, I explored ChatGPT's potential as a slot director. I posed a simple operational question: “What is the best percentage of Participation or Lease games to have on the casino floor?”

Impressively, it generated an answer over five paragraphs long in mere seconds. However, the response was quite generic. My conclusion at the time was: “Based on the non-specific output to my question, it looks like this latest version of ChatGPT relative to the casino world is at the college freshman level with little or no real-world experience.”

It's important to remember that the Chat version I used then was 3.5/4o, still in its early stages. In AI software development, "months" are akin to "years" in the real world. So, has anything changed in the subsequent 20 digital years?

Two Years On: ChatGPT-4o's Advanced Capabilities Unleashed

This time, I utilized the latest (paid) ChatGPT-4o from Open AI, leveraging its new “Deep research” feature. I also presented a much more complex question:

“How can a locals’ casino in Southern California make its slot floor more profitable?”

This is admittedly a broad query. To ChatGPT's credit, it proactively asked for more information before proceeding: “To make sure I give you the most useful advice, could you clarify a few things?”

View of a casino slot floor

It requested about a dozen details, but I only provided the casino’s square footage and reiterated my goal to increase net profitability, asking the program to focus on slot machine selection and layout optimization. This time, my desktop took significantly longer – a seemingly eternal 18 minutes!

During its process, ChatGPT consulted 28 internet sources and performed 57 individual searches. Some references included our own CDC Gaming stories, alongside information from Casino Style, Eilers & Krejcik, SAS, UNLV, and others. Interestingly, Chat even quoted me several times and referenced my editorial nemesis, UNLV’s Dr. Anthony Lucas (known for his view that most slot machines could be tighter).

Notably, Chat did not include information from many excellent books on the subject, likely due to copyright restrictions.

So, how did it perform? It's definitely no longer a freshman. I'd assess the 6,000-word essay it produced as MBA-level stuff. From my teaching experience, I’d give this paper an A+ if it came from an undergraduate. A newer Slot Supervisor could study this material and make a strong impression in a job interview to advance their career. The full output (linked below) also cited reference sources, as any good “white paper” would.

Deeper Dive: What ChatGPT-4o Uncovered for Casino Profitability

To save you some time, here are a few of the report’s suggestions:

  • Leverage analytics of actual play data to identify top-performing games and remove underperformers. Modern casinos use advanced forecasting that considers trends, seasonality, denomination, and even time-of-day usage to understand what drives a machine’s success. Rather than relying on gut feeling or historical popularity alone, data analysis can reveal which games resonate with the local market.

  • Consider offering lower minimum bets and more incremental bet options on machines to appeal to players who budget their play. A casual local patron might prefer a game where they can bet just $0.50 or $1 at a time and play longer, rather than being forced into a $3 minimum bet on a penny video slot.

  • Instead of long, endless rows, today’s floors use clusters, pods, and curving pathways to break up sightlines and invite customers to wander. This encourages a meandering flow where players naturally stroll through different sections of the floor, increasing the chances they’ll spot a machine that catches their interest.

  • The success of Station Casinos underscores that in a locals’ market, reputation and player trust in slot fairness is crucial. By not pushing hold percentages to the max and constantly updating the gaming mix (they rotate new slots from all major manufacturers, often showcasing “Coming Soon” banks to create anticipation), they keep locals engaged and on property.

  • Giving players enough personal space and comfort at the machines has proven to increase time-on-device and revenue. Pre-2020, many casinos packed machines tightly – it was common to see slim stools crammed together with barely a couple of feet between players, all in the name of maximizing units on the floor.

These are just five of the dozens of suggestions spanning 15 pages. (If you have nothing better to do for the next 25 minutes, you can read the entire output here). It would be rare to find anyone who would disagree with the concepts Chat cited. But, as any decent slot director knows, the devil is in the details.

The AI Evolution: From Data Silos to Intelligent Transformers

Before the year 2000, those “details” were primarily derived from years of experience and considerable, costly trial and error. Then came powerful “information warehouses” and illuminating “data visualizations,” which greatly improved things.

However, the real Artificial Intelligence revolution has been gaining significant momentum recently. In 2017, GPT or Generative Pre-training Transformer technology began to emerge from Open AI and others. This was like having the “Lord of the Internet” at your disposal, using a supercomputer to make sense of it all. We can now query slot machine performance against player demographics, engagement patterns, time-on-device, preferences, devotion, and promotional effectiveness. Admittedly, we could always do this, but it was often impractical due to the weeks or months it would take. Now, answers can arrive in minutes.

And it’s getting faster. This last decade is also known as the age of machine learning. This allowed AI programs to perform their own version of “trial and error,” constantly improving themselves at what seems like lightspeed.

A Word of Caution: The Imperfections of Internet-Reliant AI

Buddy Frank author photo

Make no mistake, ChatGPT-4o is not perfect. Its biggest shortcoming is that its research foundation is the internet – the same source that brought us the “Tide Pod Challenge” and theories about “fake moon landings.” I could write a book about the false and misleading garbage online, but Walter Scheirer from the University of Notre Dame has already done so.

Nevertheless, the improvement in the credibility and capability of AI and GPT technology in just a few years has been remarkable. Current AI-based analytic software in our casino industry is also improving at astonishing rates.

The Competitive AI Landscape and Its Broader Impact

Competition ensures this trend will continue. Just observe the GPT developments emerging from all corners: Apple Intelligence, Chatsonic, Claude, CoPilot, DeepSeek R1, Ernie, Gemini (formerly Bard), Jarvis (not just an Ironman tool), Jasper, Llama-4, Perplexity.ai, and dozens of others.

Some gaming regulators are now using these tools to craft new tech guidelines. CEOs are delegating administrative tasks to focus more on their core missions. IT coders are becoming faster and more efficient. So, unless you are concerned about “Skynet,” it is an exciting and potentially rewarding time for those who embrace the future.

The Verdict: Is ChatGPT Ready to Take Over Slot Direction?

So, is ChatGPT now a better slot director? The answer is clearly “Yes,” but thankfully not yet to the point of replacing us, despite our comparatively slow and methodical nature. Along with that encouraging news, being aggressive and embracing the tools created by AI and GPT will make us better equipped to withstand any assault. Conversely, failing to keep up could mean that your potential replacement is coming much sooner than I anticipated in September of 2023. Check back soon.


Postscripts: Peeking into an Even Smarter AI Future

Postscript #1: I asked if another innovative new program called Agentic AI, working in tandem with ChatGPT-4o, might perform even better as a slot director:

“Agentic AI would give ChatGPT the ability to take initiative, make decisions, and act independently—like a virtual assistant with real responsibility. As a slot director, it could continuously monitor floor performance, identify underperforming machines, and recommend real-time changes without waiting for human input. It could autonomously run test configurations, analyze player behavior, and adjust machine placement or themes to boost engagement and profit. Instead of just answering questions, ChatGPT would function like a tireless, data-driven slot manager—learning from results and acting proactively to optimize the casino floor every day. It becomes a partner, not just a tool.”

Postscript #2: Then I asked ChatGPT to produce a shorter version of the main article above. In less than two seconds, it generated the following:

“In 2023, Buddy Frank tested ChatGPT-3.5 on a casino operations question and found its response too generic, likening its understanding to that of a college freshman. Two years later, using ChatGPT-4o with its advanced “Deep research” feature, he posed a far more complex question about maximizing profitability at a locals’ casino and received a highly detailed, MBA-level analysis backed by extensive research. While AI hasn’t yet replaced the slot director, Frank now sees it as a fast-evolving tool capable of generating actionable, data-driven insights that would have once taken weeks to compile.”

Clearly, GPT won’t be a hit with freelance writers who get paid by the word.


Read Original Post
ImaginePro newsletter

Subscribe to our newsletter!

Subscribe to our newsletter to get the latest news and designs.