Casinos Limit AI to Back-Office Systems Over Licensing Risks

因担忧牌照风险 赌场将AI限制在后台系统 严格的监管审查以及对黑箱错误的零容忍，迫使赌场运营商放弃面向客户的自主智能体，转而采用可审计的运营工具。 专家组详细阐述了预计到2030年规模将达1万亿美元的全球博彩业，如何在打破传统数据孤岛的同时，积极在后台场地优化、预测性反洗钱及实时玩家个性化服务中部署AI。企业对自主智能体AI的采用依然极为审慎，因为监管层面的失误可能导致博彩牌照被吊销，且部落博彩更看重社区就业而非传统的削减成本。演讲嘉宾强调，博彩业严格的合规标准甚至超越了FDA的医疗审批，使该行业成为严谨数据治理的典型范例。展望未来，专家们预测行业将迎来双轨发展模式：前端结合预测性VIP虚拟形象礼宾服务，后台实现一键式自主营销活动编排。 全球博彩业年收入已超过7,000亿美元并有望在2030年突破1万亿美元；UNLV国际博彩研究所与毕马威联合发布的报告显示，赌场运营商正加速打破酒店、餐饮与游戏场域的数据孤岛，构建全渠道客户档案。 (2:55) 目前AI在博彩领域的应用主要集中在后台IT、网络安全与游戏研发，同时运营商也在积极利用AI进行赌场楼层优化、老虎机动态定价以及预测性反洗钱（AML）监测。 (13:02) 自主式Agentic AI在博彩企业落地面临严峻挑战，因黑盒决策失误可能直接导致博彩牌照被吊销；跨行业数据显示仅有7%的Agentic AI试点能进入实际生产，审慎部署反映了行业极高的合规标准而非技术落后。 (19:44) 博彩业的合规认证严苛程度已看齐甚至超越FDA医疗标准，2026年的监管与运营重心正从模型本身转向数据治理与隐私隔离，未来场景将覆盖个性化VIP虚拟管家与自动化后台运营。 (25:22)

Industry executives Kiran Brahmandam and Earle Hall, alongside UNLV researcher Dr. Kasra Ghaharian, argue that while the gambling industry commands an expansive quantitative footprint, AI deployment is fundamentally constrained by strict regulatory compliance, deterministic safety requirements, and the existential risk of license revocation. Consequently, gaming organizations prioritize auditable back-of-house automation and behavioral modeling over experimental, black-box autonomous agents. High-Stakes Regulation and Deep Data Footprints (2:55) Contrary to vintage cinematic depictions of gut-driven operations, the commercial gambling sector generates over $700 billion annually and is projected to surpass $1 trillion by 2030. Panelists emphasize that the modern gaming floor constitutes an extraordinarily granular consumer data environment covering hotel stays, dining, live entertainment, and gaming transactions. Earle Hall notes that casino oversight involves extensive personal background and financial checks, subjecting operators to regulatory hurdles that he argues exceed medical device certifications under the FDA. Because non-compliance or unexplainable algorithmic anomalies can result in the immediate forfeiture of gaming licenses, operators cannot risk deploying unvalidated systems. Brahmandam points out that tier-one suppliers invest heavily in self-regulation, early problem gambling detection algorithms, and research partnerships with institutions like UNLV, contrasting this accountability with mainstream tech platforms that monetize unmonitored user engagement. Legacy Silos and Back-of-House Deployment (13:02) According to the UNLV-KPMG State of AI in Gaming report—which benchmarked more than 80 gaming organizations—AI adoption is heavily concentrated in back-of-house operations, game design, IT, and cybersecurity rather than customer-facing interfaces. Brahmandam explains that historic profitability previously insulated casino operators from the margin pressures that forced early IT overhauls in retail, allowing 10-to-20-year-old legacy systems to persist. However, expanding customer touchpoints across digital apps and physical properties have recently forced operators to integrate these fragmented data environments. Regulatory realities further divide deployment strategies. Core game mechanics and slot machine math face rigid certification barriers under technical standards like GLI, restricting front-of-house generative experimentation. In contrast, back-office casino management systems enjoy substantial latitude to implement AI for pricing, floor placement, predictive anti-money laundering (AML), and marketing allocation across roughly 140 properties serviced by Gaming Analytics. Auditability Hurdles for Autonomous Agentic AI (19:44) The UNLV maturity index revealed that organizational ambition currently outpaces technical infrastructure, expertise, and governance across the gaming sector, particularly regarding autonomous agentic AI. Brahmandam stresses that while consumer agents perform low-stakes functions like email summarization, a casino delegating a $20 million operational budget to an autonomous agent requires absolute auditability, deterministic consistency, and structured context layers built on proprietary knowledge graphs. General vendor pitches emphasizing payroll reduction also fail within Tribal gaming operations, where mandates prioritize Native community employment and social investment over operational cost-cutting. Hall argues that the industry's cautious adoption reflects prudent enterprise risk management rather than technological backwardness. Citing cross-industry benchmarking data from an AWS executive briefing, Hall highlights that while 85% of agentic AI initiatives reach the pilot stage, only 7% successfully transition into enterprise production across all commercial sectors. Dynamic Player Modeling and Automated Orchestration (32:04) Audience discussion highlighted a major limitation in traditional casino marketing: the reliance on arbitrary 90-, 180-, or 365-day inactivity buckets to classify customer churn. Brahmandam explains that static windows misallocate capital by over-subsidizing infrequent annual visitors while failing to notice abrupt departures by daily players for months. Modern AI addresses this by learning individual visitation frequencies, enabling operators to detect behavioral anomalies and intervene in real time before churn occurs. Looking three to ten years ahead, the panelists project divergent frontiers for casino AI. Hall anticipates a predictive guest concierge—a personalized VIP avatar butler integrating external services like Uber, OpenTable, and venue ticketing to orchestrate customized trips upon arrival at Harry Reid International Airport. Brahmandam envisions autonomous back-office execution where casino executives define high-level strategic targets, allowing AI agents to generate auditable operational recipes that automatically coordinate hotel bookings, direct marketing offers, and loyalty adjustments under human oversight. The panel underscores that enterprise AI adoption in gaming hinges on resolving data governance, access boundaries, and auditability rather than advancing model capabilities alone. A primary unresolved challenge remains aligning strategic ambitions with internal technical maturity, as operators seek to harness real-time predictive analytics without breaching strict deterministic compliance mandates.

在内华达大学拉斯维加斯分校（UNLV）与Ai4联合主办的产业研讨中，学者与行业高管指出，尽管全球博彩业年收入已超7000亿美元并向2030年万亿美元规模演进，但极高的合规壁垒与牌照吊销风险，迫使该行业对人工智能特别是自主智能体采取审慎态度，形成了重后台运维与确定性控制、审慎推进前端黑箱决策的非对称布局。 监管约束与数据架构演进 (2:55) 针对外界将赌场运营视为依赖直觉或洗钱温床的刻板印象，Earle Hall与Kiran Brahmandam指出，现代博彩业是受全球最严苛法律审查的量化产业，其从业背景调查严格，且博彩税收直接用于支持各州的基础设施与公共教育。现代综合度假村涵盖机台、酒店、餐饮与大型演艺，构成极其丰富的消费者行为数据流。历史上由于赌场业务利润极高，运营商缺乏动力改造沿用十至二十年的老旧系统，导致数据长期处于割裂状态；但近五至七年全渠道互通的趋势推动行业转向云原生和无现金ERP生态体系。 随着跨业务数据整合推进，数据主权与治理成为核心议题。Hall与Brahmandam强调，博彩业面临的技术审查标准甚至高于FDA对医疗设备的认证，底层游戏数学逻辑必须通过严格合规检测（如GLI标准）。因此，2026年产业的竞争焦点已从基础算法模型能力，转向确定数据驻留权、第三方访问权限以及隐私边界的数据治理架构。 后台运营主导与合规诉求分化 (13:02) UNLV国际博彩研究所与KPMG联合发布的《博彩业AI现状》报告（调研超80家机构的核心管理层）显示，AI落地高度集中于后台运营、游戏产品研发、IT基础架构及网络安全，而直接面向玩家的前端互动与合规风险控制领域的采纳度相对较低。Brahmandam以Gaming Analytics服务约140家赌场（约占市场15%）的经验解释，在实体运营端，AI已被广泛用于老虎机机位布局、机台采购评估、动态定价优化、实时营销推荐以及反欺诈反洗钱检测。 在合规风控领域，AI的应用重点呈现明显的区域分化。Hall分析指出，预测性反洗钱（AML）已成为全球赌场的标准化通用底线，而在算法实现上相对容易；相比之下，澳大利亚、英国与新加坡等监管环境极为严格的市场，对“预测性危害最小化”（harm minimization）展现出强烈需求，要求算法能实时识别玩家是否偏离正常下注节奏并进入危险博弈状态，以履行更严格的负责任博彩义务。 牌照红线与确定性诉求制约智能体 (19:44) 针对博彩业在自主智能体（Agentic AI）应用上的明显滞后，Brahmandam指出企业级高风险场景与消费级AI存在本质差异。个人智能体处理日程汇总或酒店预订时具有容错空间，但赌场若将数千万美元的营销或调度决策托付给黑箱算法，一旦产生不可解释的错误，不仅会造成直接资金损失，更会导致博彩执照被直接吊销。博彩监管机制要求任何算法决策必须具备绝对的确定性与可审计性，同一问题在受审时必须输出完全一致、可验证的推理路径，这使得缺乏专有上下文层与知识图谱的通用大模型无法直接进场。 产业的审慎立场同时反映了特殊的商业结构与全行业转化瓶颈。Brahmandam强调，在部落博彩（Tribal Gaming）体系中，企业的首要目标是保障部落成员就业与社区社会投资，而非采纳外部技术供应商倡导的激进裁员降本逻辑。Hall援引AWS闭门研讨数据佐证，跨行业来看，全美85%的智能体项目虽能进入试点，但最终仅有7%能转化为实际生产力。UNLV成熟度指数显示，尽管博彩业管理层的战略雄心已大幅超出其基础设施和治理准备度，但在真金白银的交易场景下，谨慎部署恰恰避免了对玩家进行不负责任的算法实验。 动态频次建模与人机协同愿景 (32:04) 在解决客户流失与营销效率问题上，传统赌场长期依赖静态时间切片（将90天、180天或365天未到访者机械划分为流失客），导致低频年访客被过度投入营销资源，而突发中断的高频常客却在长达一年内无法被系统预警。Brahmandam指出，AI的核心价值在于实时构建单客独特的行为频次基线，在行为模式破裂的最初节点自动发起干预，实现全生命周期的前置维护，消除粗暴的时间切片弊端。 关于未来三至十年的图景，两位高管提出了人机协同与体验重塑的双重构想。Hall描绘了前端客户旅程的演化：未来游客在抵达拉斯维加斯前便可接入专属数字化VIP管家，依托集成OpenTable、Uber及票务生态的预测性大模型，在碎片时间内自主推荐并预订Sphere球体场馆等高端体验。Brahmandam则勾勒了后台运营的“自主编排”模式：管理层仅需输入业务指标（如月收入增长3%），智能体系统即刻推演出多套执行策略配方，在经由管理者人工微调校准后，一键自主调度房态、推送个性化优惠短信并同步忠诚度系统，达成受控且闭环的高效运营。 博彩行业对AI的采纳表明，高资本收益与严苛牌照红线共存的行业中，技术落地受制于审计合规而非模型能力。尽管UNLV数据显示博彩企业的战略野心与其IT基础设施和数据治理能力之间尚存显著断层，但在可解释性与确定性获得监管认可之前，以人为核心控制回路的后台编排与动态生命周期建模，仍是该产业平衡创新与风控的主要路径。

## Background 本内容由 Ai4 于 2026 年 8 月 14 日发布在 YouTube 平台上。视频记录了内华达大学拉斯维加斯分校（UNLV）国际博彩研究院研究总监 Kasra Ghaharian、Aristocrat 旗下 Gaming Analytics 董事总经理 Kiran Brahmandam 以及 AXES.ai 总裁兼首席执行官 Earle Hall 共同参与的行业研讨会，重点基于《2026 年游戏领域 AI 现状报告》探讨博彩与实体娱乐场行业在 AI 部署、合规监管、数据治理与未来运营模式上的核心挑战与前沿变革。 ## TL;DR 本场座谈深入破除了大众对博彩娱乐业依靠直觉运转的刻板印象，指出该产业实则是受全球最严苛合规与安全审查、沉淀超高维消费者行为数据的科技密集型领域。嘉宾们结合 UNLV 最新行业调研指出，博彩业在采纳自主智能体（Agentic AI）上面临极高的牌照与财务犯错成本，因而相较通用商业领域展现出更审慎的负责任 AI 态度；目前 AI 应用主要集中于后台支持、机台调优与预测性反洗钱。展望未来，博彩业将迈向由 AI 虚拟管家重塑玩家全旅程体验，以及由管理层设定顶层商业目标、智能体端到端自主执行营销与运营协同的新范式。 ## Atmosphere & style 整场研讨气氛专业严谨却又不失赌城特有的幽默与生动，三位常驻拉斯维加斯的行业领袖在台上默契十足，谈吐间兼具学术研究的严谨与一线实战的敏锐。开场对拉斯维加斯酷热与人潮的调侃迅速拉近了与跨界技术观众的距离，随后关于高昂犯错成本、牌照风控及玩家生命周期模型的深度剖析则展现出极强的行业纵深。现场既有对主流 AI 大会中博彩议题边缘化的风趣自嘲，更有对数据治理与伦理底线的坚定坚守，让观众深切感受到博彩科技在拥抱 AI 浪潮时那份独有的克制与硬核。 ## Key takeaways - [02:55](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=175s) 博彩业并非凭直觉运转的江湖行业，而是全球年营收超 7000 亿美元且受最严苛监管的庞大经济体；其涵盖博彩、酒店、餐饮与文娱的高维并发数据，为 AI 提供了极具价值的沉淀土壤。 - [06:26](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=386s) UNLV 国际博彩研究院发布《游戏领域 AI 现状报告》，将 AI 重塑现代赌场的核心逻辑提炼为“数据（Data）、安全（Safety）与体验（Experience）”三大支柱。 - [08:12](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=492s) 博彩业长期因丰厚现金流与盈利壁垒对技术革新较为保守，但年轻客群全渠道、跨平台的娱乐习惯正在倒逼赌场打破数十套“数据孤岛”，加速拥抱智能化转型。 - [11:42](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=702s) UNLV 与毕马威的调研显示，当前博彩业 AI 落地高度集中在技术研发、网络安全与产品创新等后台职能（Back of House），而在前台客户交互与合规风控方面的应用相对滞后。 - [14:25](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=865s) 约 15% 的赌场市场已在楼面机台选型、动态定价及实时营销中接入 AI，但全行业在系统化挖掘长期“玩家旅程”（Customer Journey）与合理评估客户价值方面仍存在认知断层。 - [18:44](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1124s) 智能体 AI（Agentic AI）在博彩业落地滞后的根本原因在于极高的犯错成本——决策失误可能直接导致博彩牌照被吊销，且北美部落赌场排斥以裁员降本为导向的 AI 叙事。 - [23:05](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1385s) 全行业仅 7% 的 Agentic 项目能进入生产环境（85% 止步于试点）；博彩业涉及真实真金白银与玩家福祉，在 AI 部署上的审慎与负责任态度反而是保护玩家的行业优势。 - [25:22](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1522s) 2026 年博彩业的核心议题正从算法转向底层数据治理（存储、权限与流向），同时行业正加大对责任博彩（Responsible Gaming）和成瘾早期识别算法的投入以应对监管关切。 - [30:33](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1833s) 严格的行业监管并未扼杀后台运营决策类 AI 的创新，但在物理老虎机研发与发牌认证上构成了极高的准入门槛与竞争壁垒。 - [31:56](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1916s) 未来博彩娱乐体验将由预测型 LLM 虚拟管家重塑，实现交通、订餐与娱乐的无缝调度；后台管理者则可通过自然语言直接下达商业目标，由系统自主拆解策略。 - [35:17](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=2117s) 智能体驱动的运营新范式确立为“管理者设定顶层目标并根据经验微调，AI 与自动化软件全渠道端到端执行”，大幅缩短营销策划与资源调度周期。 - [36:35](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=2195s) 传统赌场依赖 90 天或 365 天等固定时间窗口来判定玩家流失存在巨大误区，AI 的真正价值在于建立千人千面的动态基准，在玩家偏离个体行为节律的第一时间实现精准预警与留存。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=4s) 欢迎致辞与嘉宾阵容介绍 博彩与人工智能领域的年度峰会拉开帷幕，内华达大学拉斯维加斯分校国际博彩研究院（UNLV International Gaming Institute）的研究总监 Kasra Ghaharian 博士登台主持。他首先向观众致谢，并打趣道研究院其实就在会场向东仅 10 分钟车程的地方，站在拉斯维加斯大道（The Strip）上甚至清晰可见。谈及这场 AI4 行业大会，Ghaharian 回忆起自己在 2023 年于美高梅大酒店（MGM）首次参会的情景，直言当年的盛会就已“人满为患”，而如今的热度更是空前。他甚至分享了前一天的亲身惊险遭遇：搭乘扶梯到达顶端时，人群突然在出口处完全停滞堵塞，险些酿成踩踏意外，万幸的是主办方随后立刻安排了工作人员在电梯口疏导人流。 面对会场琳琅满目的分论坛选择，Ghaharian 对在场观众把未来 40 分钟留给博彩 AI 议题表示感谢，并郑重请出台上的两位重磅行业领袖进行不超过 60 秒的自我介绍： * **Kiran Brahmandam**：现任知名博彩巨头 Aristocrat 的 Gaming Analytics 董事总经理。他幽默地向后排观众确认声音是否清晰，并介绍 Aristocrat 是一家业务体量庞大的综合游戏公司，不仅生产了铺满各大实体赌场的知名老虎机（如风靡全美的旗舰产品“Buffalo”），其业务版图还覆盖了线上娱乐平台与社交类游戏开发。在加入该公司前，他创立了深耕博彩 AI 运营变革长达 8 年的初创公司 Gaming Analytics，该公司近期正式被 Aristocrat 成功收购。 * **Earle Hall**：AXES.ai 总裁兼首席执行官。他一开场便幽默地调侃赌城室外的滚滚热浪，祝大家享受室内清凉的冷气。Hall 指出，台上的三位专家均常驻拉斯维加斯，多年来深度参与博彩行业标准制定与监管咨询；而他的日常核心工作是带领团队研发一套集“云端、无现金化与 AI 驱动”于一体的企业资源规划（ERP）系统，专为驱动实体地面博彩场所的数字化运营而设计，同时他还兼任区块链等多项行业标准委员会的委员职务。 --- ### [02:55](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=175s) 破除行业刻板印象：监管深度与超高维度消费数据 针对大众对博彩娱乐业的普遍认知，Ghaharian 博士提出破题探讨：许多外行人对该领域的想象仍停留在数十年前的影视镜头里，诸如《赌城风云》（*Casino*）或《十一罗汉》（*Ocean's Eleven*）中展现的那样，认为这只是一个全凭直觉和运气运转、毫无科学性可言的江湖行业。然而现实大相径庭——博彩业如今是一个高度由数据驱动的庞大经济体，全球年营收已突破 7000 亿美元，并预计在 2030 年跨过 1 万亿美元大关，这为 AI 技术的规模化渗透孕育了极具吸引力的沃土。 谈到外界最大的误解时，Kiran 指出，大众不仅常把垂直领域的 Gambling 与一般广义的 Gaming 混淆，更普遍低估了该行业的科技密集度以及合规治理成本。外界习惯于各类社交媒体上的“无限刷屏”（doom scrolling）与消费诱导，却很少意识到现代博彩业在自我监管、合规运营和保护玩家健康方面倾注了何等惊人的资金、精力和技术。这个行业极其严苛地践行“负责任博彩”，而其产生的税收更是切实反哺了各州的公立教育体系和交通基础设施建设。 来自加拿大魁北克的 Earle 则从道德与社会认知的角度直言不讳：外界常常将博彩等同于洗钱与上瘾等负面标签，但现实中博彩是全球受到最严酷、最过度监管（overregulated）的行业之一。从业者进入该体系不仅要上交个人银行账户、全量信用卡账单，监管机构甚至会逐一走访其邻居以审查个人操守。不仅如此，很多人未曾洞悉的是，博彩底层承载着当代商业中最深厚、最复杂的高维消费者体验。一个踏入综合度假村赌场的顾客，可能同时在预订顶级餐厅、延长酒店住宿，或是安排前往沉浸式球体场馆 Sphere 与观看太阳马戏团（Cirque du Soleil）的演出。这一系列并发轨迹构建出了消费零售领域最具丰富度与全息维度的“用户画像”（Avatar），为利用 AI 赋能玩家体验带来了无限施展空间。 ```mermaid flowchart TD accTitle: 现代综合度假村多维消费者数据流 accDescription: 展现玩家在赌场空间内产生的交错并发消费数据，汇集为高维零售用户画像。 subgraph User["核心消费主体"] Player["娱乐场玩家 (高维数据 Avatar)"] end subgraph Activities["并发行为触点"] direction TB A1["现场博彩活动<br/>(老虎机 / 桌台流水)"] A2["酒店款待服务<br/>(客房预订与延住)"] A3["高档餐饮消费<br/>(餐厅预约 / 消费偏好)"] A4["现场文娱演出<br/>(Sphere / 太阳马戏团门票)"] end subgraph DataEngine["数据与监管引擎"] CMS["实时 ERP 与云端分析系统"] RegEngine["严格合规与负责任博彩监控"] end Player --> A1 Player --> A2 Player --> A3 Player --> A4 A1 & A2 & A3 & A4 --> CMS CMS <--> RegEngine ``` --- ### [06:26](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=386s) 行业里程碑发布：2026 年《游戏领域 AI 现状》行业报告 Ghaharian 博士表示，本次演讲团队习惯了在专业博彩产业大会上发言，而面对 AI4 这样汇聚跨行业技术精英的舞台，他们特意做了精心的针对性筹备。为了让跨界从业者在有限时间内把握行业脉搏，本次研讨被精炼聚焦为三大核心命题：**数据（Data）、安全（Safety）与体验（Experience）**。现场快速举手表决显示，台下既有深谙线下赌场的从业者，也有专注算法研发与架构落地的技术主管。面对两类人群交融的观众席，主持人打趣称“刚才我们唯独没有统计现场有多少人自己平时爱下注博彩”，引发现场心领神会的欢笑。 随后，大屏幕展示了 UNLV 国际博彩研究院于同年 4 月份重磅发布的《游戏领域 AI 现状报告》（*State of AI in Gaming Report*）。该份研究长达百余页，凝聚了学术界与产业界的最新实证数据。尽管 Ghaharian 表现得十分谦逊，但身旁的 Earle 当场盛赞这份报告，直言自己已将其称为该垂直赛道的“圣经”，甚至是“迄今为止跨所有行业阅读过的最佳 AI 研报之一”。该报告向全行业免费公开开放，并作为本次前沿对谈的核心数据依托与逻辑基底贯穿始终。 --- ### [08:12](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=492s) 技术演进之困：暴利壁垒、数据孤岛与跨渠道整合 在展开论述第一个核心支柱“数据”时，主持团队指出，赌场内部沉淀着整个消费社会中极其庞大、却几乎未被外部 AI 开发者真正谈论过的富矿级数据集。然而，赌场底层的数字化路径并非一蹴而就： * **20 世纪末（前 2000 年代）**：仍属于机械时代，以投币式、物理拉杆的传统机械老虎机（One-armed bandits）为主体，几乎不具备实时数字追踪能力。 * **1990 年代末至 2000 年代初**：迎来电子化浪潮，引入 TITO 凭证系统（Ticket-In, Ticket-Out 电子出钞技术）与早期的娱乐场管理系统（CMS），逐步将楼层活动转化为数字日志并开启会员卡跟踪。 * **当下时代**：物理实体空间与移动数字端趋于融合，赌场底面终于迎来了所谓“数据就绪”（data ready）的 AI 应用时代。 然而在探讨这种技术转型是否平滑顺畅时，Kiran 直截了当地打破了所谓“完美演进”的幻想。他坦言博彩业与外部传统行业一样，长期被严重的“数据孤岛”所困扰，各类交易记录、会员积分、行为监控常常分散在数十乃至上百套互不联通的老旧孤立系统之中，管理人员很难统揽全局作出基于全局视角的即时决策。 更具戏剧性的是，该行业面临着一个特有的结构性矛盾——“这行业的好处是赌场极其赚钱，而坏处也是赌场太赚钱了”。在外界零售业分厘必争、疯狂压缩成本以换取微薄利润率的环境下，实体赌场坐拥丰厚的现金流护城河，运营层极易产生惯性思维：“既然过去 10 年、20 年所沿用的旧流程就能赚得盆满钵满，我们为什么非要自讨苦吃去推动技术变革？”博彩业曾长期享有任由外部科技浪潮演进数年、自己稍后再慢节奏跟进的特权，正如 Ghaharian 的导师曾精辟评价的那样：“这个行业哪怕逆自身运营逻辑而行，也照样日进斗金。” 然而，Kiran 敏锐地指出，这种保守心态在过去 5 到 7 年间发生了剧烈扭转。年轻客群不再单纯守在物理机台或牌桌前，他们的注意力延伸到了移动互联、线上投注以及社交娱乐游戏之中。玩家行为从物理单点扩展到了全渠道、多平台，促使赌场运营商急迫地需要把所有触点数据编织收拢，以此打磨前所未有的无缝沉浸式体验。这也正是如今该产业对具身与自主智能体（Agentic AI）展现出前所未有热忱的根本动力。顺接这一话题，Earle 也重申了赌场在本质上极度“渴求数据”的天然属性，现代博彩系统正在不遗余力地实时探查每一秒驻留在场内的用户动向、情绪安全与合规边界。 ### [11:42](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=702s) 博彩业的数据渴望与行业报告揭示的 AI 应用现状 博彩业在本质上是一个对数据极度渴望的行业。去赌场并非为了赚钱——如果有人能带着钱离开，那只是百万分之一的幸运儿，玩家真正购买的是类似于去高档餐厅消费的娱乐体验。为了保障安全与娱乐并存，行业需要打破餐饮、酒店、桌面游戏（Table Games）等传统数据孤岛，深入了解玩家身份与游戏状态。加之反洗钱（AML）和责任博彩（Responsible Gaming）等强监管要求，博彩业在基因上就具备了高度的数据驱动特性。 ```mermaid flowchart TD accTitle: 博彩业数据驱动与AI应用全景 subgraph S1["数据孤岛打通"] A["餐饮业务 (F&B)"] B["酒店住宿 (Hotels)"] C["博彩娱乐 (Table Games & Slots)"] end subgraph S2["监管与业务诉求"] D["反洗钱合规 (AML)"] E["责任博彩 (Responsible Gaming)"] F["玩家旅程洞察 (Customer Journey)"] end subgraph S3["AI 落地场景"] G["后台技术与安全运营 (Back of House)"] H["前台赌场楼面优化 (Floor Operations)"] end A & B & C --> D & E & F D & E & F --> G & H ``` UNLV 国际博彩研究院（International Gaming Institute）联合 KPMG 对 80 多家博彩相关机构展开了深度调研。为了保证数据质量，每个组织仅限一名高管受访（受访者涵盖 CEO、首席 AI 官 CAIO、CTO 等层级）。调研呈现出一个引人深思的现象：当前的 AI 实际用例高度集中在技术、安全、产品开发与创新等后台支持职能（Back of House），而在面向客户的前台功能（Customer-facing）以及风险合规职能中的应用反而相对滞后。 - **调研样本规模**：覆盖超过 80 家博彩企业，采取单一组织单份高管答卷的高标准数据采集。 - **后台集中趋势**：技术研发、网络安全与产品创新构成了当前博彩业 AI 落地的主阵地。 - **前后台落差**：前台客户体验与合规风控方面的 AI 直接落地比例低于后台技术开发。 --- ### [14:25](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=865s) 供应商与运营商的落差：从楼面决策到理解“玩家旅程” 针对调查中前后端的应用差异，Kiran Brahmandam 指出受访者构成的偏差可能是原因之一：软件与硬件供应商自然会将 AI 资源倾斜在代码编写、技术架构和安全开发上；但就实际运营商而言，过去四到五年间，AI 在赌场楼面运营（Casino Floor Management）中已经发挥了显著作用。以 Gaming Analytics 服务的约 140 家赌场（约占市场份额的 15%）为例，运营商广泛利用 AI 决定老虎机采购、机台摆放位置、动态定价策略、营销活动调整、实时玩家个性化推荐以及反欺诈检测。 Earle Hall 补充表示，AI 的核心诉求往往存在地域文化差异：在澳大利亚、英国和新加坡等地，监管机构与运营商非常关注利用数据预测玩家是否超出了财务承受范围（进入危险体验区间），而算法层面最容易落地的则是预测性反洗钱。但全球运营商最根本的共同困惑始终是：“谁是我们的玩家？他们何时来、为何来、停留多久，又为何突然离场？” 博彩业长期存在一个痛点——“它是唯一一个不太懂得如何给客户合理估值的行业”。尽管业界提出了“玩家旅程”（Customer Journey）的概念，但在先进技术方案与客户认知之间依然横亘着巨大的教育鸿沟。相较于星巴克或宝马通过 App 建立长期客户关系的模式，博彩业手握远比普通零售业丰富的数据资产，却仍在摸索如何将其转化为长期的客户价值与更安全的娱乐互动。 - **市场落地现状**：约 15% 的赌场市场已在楼面运营决策、老虎机选型定价与实时优惠发放中接入 AI 工具。 - **区域合规关切**：澳大利亚与欧洲重在防范过度博彩与不健康消费行为，预测性反洗钱则是全球最成熟的算法需求之一。 - **玩家旅程认知断层**：赌场急需理解玩家在单次下注（Session）、单次到访（Visit）以及长期生命周期中的动态行为，但多数机构缺乏系统化挖掘能力。 --- ### [18:44](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1124s) 智能体 AI（Agentic AI）的落地瓶颈与博彩业的独特代价 在细分技术应用调研中，最突出的结果是博彩行业对智能体 AI（Agentic AI）的采用极其滞后，明显落后于其他商业领域。Kiran 分析了其中的底层逻辑：在个人消费场景下，AI 智能体处理日常邮件或预订酒店即便出错也无伤大雅；但在高监管的博彩业，任何业务决策都要求极高的可验证性、可靠性与审计合规性。 若将 2000 万美元的营销预算全权交由智能体自动分发优惠活动，一旦决策失误且缺乏可解释性，赌场面临的不仅是资金亏损，更可能是博彩牌照被吊销的灭顶之灾。这种“犯错成本（Cost of a Mistake）”的高昂程度堪比医疗行业。当前 Agent 架构仍需依赖专门的组织级数据上下文层（Context Layer）、知识图谱（Knowledge Graph）与子图查询等复杂技术支撑，尚未完全成熟到足以承担重大经营决策。 ```mermaid flowchart TD accTitle: 智能体 AI 在普通行业与博彩业的应用对比 subgraph Generic["通用/消费级场景"] A["邮件整理 / 酒店预订"] --> B["低犯错成本"] B --> C["快速采纳 Agentic AI"] end subgraph Gaming["博彩行业特殊约束"] D["重大营销预算 / 牌照合规"] --> E["极高犯错成本（吊销执照风险）"] E --> F["强需求：100% 可审计、可解释、高精度"] F --> G["Agentic AI 采用相对滞后"] end ``` 此外，通用商业领域常见的“降本增效/裁员”说辞在博彩业并不适用。许多原住民部落赌场（Tribal Casinos）的核心宗旨是造福部落社区、创造并保留就业岗位，单纯以削减人力成本为导向的 AI 方案很难被管理层接受。外部技术供应商若想立足博彩业，必须保证系统在相同输入下每次都能输出完全一致、可审计的结果，否则将直接被监管机构取缔。 调研由此构建了一个涵盖“战略（Strategy）、基础设施（Infrastructure）、专业技能（Expertise）、治理规范（Governance）”四大维度的行业成熟度指数（Maturity Index），结果清晰显示：博彩业对 AI 的宏伟雄心已大幅超越其实际的基础设施、人才储备与治理水平。 - **容错率极低**：不同于常规自动化，博彩决策失误面临撤销牌照风险，导致对自主 Agent 的采纳极为审慎。 - **部落赌场特质**：北美部落博彩注重本土就业保障与社区投资，排斥单纯以裁员降本为目标的 AI 叙事。 - **四大评估维度失衡**：在“战略、基建、人才、治理”四大支柱中，企业的战略预期与野心明显跑在基建与风控治理前面。 ### [23:05](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1385s) 游戏行业在 AI 落地上的“滞后”是劣势还是优势？ 针对博彩及游戏行业在基础设施、技术专长和治理结构上是否落后于其他领域的疑问，Earle Hall 分享了当天早晨 AWS 闭门早餐会上的最新关键数据：跨所有行业来看，只有 7% 的 Agentic（智能体）项目最终能够成功进入生产环境，而高达 85% 的项目仅仅止步于试点（Pilot）阶段。Earle 指出，博彩业本身处于极高强度的监管与自我监管环境中。当一家赌场想要尝试 Agentic AI 项目并向监管机构或州立博彩委员会咨询时，监管方往往也因技术的崭新性而面面相觑、无从判断。 Earle 强调，赌场现场流动的是真实的真金白银，关乎玩家的生计与真实娱乐体验，因此该行业的审慎程度远超其他任何领域。行业不仅在谨慎创新，更在践行负责任 AI（Responsible AI）与伦理标准，绝不会拿玩家当实验品。因此，如果说博彩业在 AI 部署节奏上落后于其他领域，Earle 坚定地认为“这绝对是一件好事”。 - **全行业 Agentic AI 落地率**：仅 7% 的智能体项目能真正进入生产环境，85% 的项目止步于概念验证或试点阶段。 - **审慎底线**：游戏行业因涉及真实资金与玩家福祉，在监管审核与伦理上必须坚持不以玩家为试验对象。 ```mermaid flowchart LR accTitle: Agentic AI 落地漏斗与行业审慎机制 subgraph Funnel["全行业 Agentic AI 落地现状 (AWS 数据)"] A["85% 试点阶段<br/>(Pilots)"] --> B["7% 正式投产<br/>(Production)"] end subgraph GamingConstraint["博彩行业核心约束"] C["涉及真金白银与玩家生计"] --> D["严格外部监管与行业自律"] D --> E["审慎推进 / 拒绝拿玩家做实验"] end ``` --- ### [25:22](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1522s) AI 安全、监管期待与游戏业的合规底色 主持人 Kasra Ghaharian 转向本次讨论的第二个核心议题——AI 安全（AI Safety）。虽然当前科技界都在热议 AI 安全与对齐，但博彩业自诞生数十年来就一直处于反欺诈、反洗钱（AML）以及预防赌博成瘾等高压监管的审视之下。UNLV 的最新调研报告显示，大多数监管机构认为博彩业需要比通用 AI 更严格的专属 AI 监管规则，并且对运营商和供应商的“自我监管”能力信心不足。 面对监管机构的怀疑态度，Kiran Brahmandam 从 Aristocrat 的实践出发做出了回应。他指出，责任博彩（Responsible Gaming）是 Aristocrat 的核心战略支柱，公司不仅投入巨资用于玩家教育、线下自我隔离（Self-exclusion）工具以及赌博成瘾早期识别系统的研发，还资助了 UNLV 等多所大学开展前沿研究。Kiran 进一步指出，监管机构自身也需要加速学习 AI 知识，甚至应将博彩业作为向全球推广安全 AI 实施的试验田。 Earle 补充道，博彩业的受监管程度甚至远超医疗科技及 FDA 认证。正因为受管制的合规企业本就遵循着全球最严苛的行为准则，AI 在该行业更像是既有合规文化下的又一个新工具。他特别强调，2026 年博彩业最核心、最前沿的讨论其实不是“AI 算法本身”，而是**数据治理（Data Discussion）**——数据究竟存放在哪里、流向何方、谁有权调阅、是否有第三方介入。 ```mermaid flowchart TD accTitle: 游戏行业 AI 安全与合规架构 subgraph RegulatoryContext["监管现状与考量"] R1["反洗钱 (AML) / 反欺诈"] R2["责任博彩 (Responsible Gaming)"] R3["严格准入许可 (如 GLI 认证)"] end subgraph AIImpact["AI 引入后的核心议题"] D1["底层数据治理<br/>(存储/权限/第三方访问)"] D2["玩家成瘾早期检测模型"] D3["后台运营与决策支持系统"] end RegulatoryContext --> AIImpact ``` --- ### [30:33](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1833s) 强监管会扼杀技术创新吗？ 针对主持人提出“OpenAI 和 Anthropic 发布模型无需向任何人报批，而博彩业的严监管是否扼杀了创新”的疑问，Kiran 区分了后台运营与前台机台研发两个不同维度： - **系统与后台运营（Casino Operations & Systems）**：在优化赌场运营和辅助管理层进行战略决策的 Agentic 解决方案上，严监管并未阻碍创新，团队拥有广阔的探索空间。 - **实体老虎机与游戏内容研发（Game Development / Slot Machines）**：因涉及严格的责任博彩规范与发牌认证门槛，创新确实受到诸多规制约束，行业外的新玩家很难随意切入，但这构成了该行业特有的双刃剑（既是门槛也是保护壁垒）。 --- ### [31:56](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1916s) 体验重塑：从“个人专属管家”到智能赌场运营 探讨最后转向对未来 3 到 10 年游戏与娱乐体验的展望。Kasra 抛出场景：“当我降落在拉斯维加斯的哈里·里德国际机场（Harry Reid International Airport）时，我会经历什么？” Earle 描绘了一个由预测型大语言模型（Predictive LLMs）驱动的个性化全景体验：过去游客依赖黄色出租车后座上的纸质优惠券指南探索城市，而在未来，从决定前往拉斯维加斯的那一刻起，系统就会为每位旅客分配一个专属的“VIP 虚拟管家（Personal Avatar VIP Butler）”。这个管家能聚合类似 OpenTable 的餐饮预订、Uber 的出行调度以及即时娱乐推荐。比如游客询问“我有两个小时空档该干什么”，管家便能瞬间安排好拉斯维加斯球体馆（Sphere）的 VIP 导览通道。Earle 提到，这绝非虚无缥缈的科幻，UNLV 的学生以及 Desert Ford Ventures 等本地机构已经在推进类似项目。 Kiran 随后从赌场后台运营的角度补充了他的构想：行业的焦点往往全放在玩家身上，却忽略了制定体验决策的管理者。他所设想的未来运营体验是——管理层只需向系统发出自然语言指令（例如：“能否帮我们在本季度实现营收额外增长 3%？”），底层的智能系统便能自动调用数据、模拟策略并输出最优运营路径。 - **未来旅客画像**：从落地甚至行前即接入深度 LLM 驱动的 AI 虚拟管家，无缝串联交通、订餐与娱乐日程。 - **后台运营演进**：管理层通过对话式 AI 直接下达商业目标，系统自主进行策略拆解与赌场楼层资源优化。 ### [35:17](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=2117s) 智能体驱动的自动化运营与行业未来愿景 演讲嘉宾深入描绘了未来赌场与博彩业的运营图景：在 Agentic AI（智能体解决方案）的辅助下，系统能够基于当月的业务数据和宏观目标自动计算出多种可行的运营“配方”（recipes）并推荐给管理层。高级主管不再需要为繁杂的执行细节耗费精力，而是扮演统筹把控的角色——就像大厨一样在推荐方案中“加入自己的盐和胡椒粉”（add their own salt and pepper），根据直觉和经验进行微调。随后，主管只需按下一个按钮，背后的集成技术便会全自动运转：促销优惠精准发放、底层多系统协同联动、个性化短信实时推送至玩家手机、配套的酒店客房预订同步锁定。 这种变革代表了博彩业乃至全球商业运作的根本演进方向：由运营者设定核心商业目标，再交由软件与 AI 自主执行落地。这种“目标驱动、机器执行”的技术范式不仅将彻底颠覆博彩业底层的业务运转模式，对于各行各业的数字化运营同样具备深远的借鉴价值。 ```mermaid flowchart TD accTitle: Agentic AI 业务执行架构流程图 accDescr: 展示从管理者设定目标到 AI 智能体生成方案、人工微调再到全渠道自动化执行的完整链路。 subgraph Decision["决策层：目标与调优"] A["设定商业运营目标\n(Operators Define Goals)"] --> B["AI 智能体生成策略配方\n(Agentic Recipes)"] B --> C["高管微调并确认\n('Add Salt & Pepper' & Push Button)"] end subgraph Execution["执行层：自动化系统联动"] C --> D["全渠道自动触达与执行"] D --> E["动态推送短信通知\n(Text Messages)"] D --> F["发放精准优惠方案\n(Targeted Offers)"] D --> G["自动关联酒店客房预订\n(Hotel Reservations)"] end ``` - **运营范式转变**：管理者的核心职责转向制定顶层商业目标（Define business goals），AI 与自动化软件负责端到端执行（Software executes），极大压缩了传统营销活动的策划与落地周期。 - **人机协同模式**：保留了人类管理者的把关权，高管通过对 AI 策略进行参数微调（“加入盐和胡椒”），兼顾了自动化的高效与人类行业经验的敏锐度。 --- ### [35:53](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=2153s) 主持人总结与拉斯维加斯现场的特殊语境 研讨环节告一段落后，主持人卡斯拉·加哈里安（Kasra Ghaharian）接管发言，在大屏幕上展示了三位嘉宾的 LinkedIn 社交主页，以便参会人员后续建立深入联系。卡斯拉在向现场听众致谢的同时，点出了一个耐人寻味且略带幽默的现实矛盾：当前这场盛会正举办于拉斯维加斯——公认的“世界娱乐之都”（entertainment capital of the world），毫无疑问也是全球博彩业的绝对核心与大本营，然而在整个 Ai4 人工智能大会的庞大日程中，博彩垂直领域的代表性却严重不足。 “我们实际上是本届大会唯一一个与博彩相关的内容板块，”卡斯拉说道。尽管身处主场，传统博彩科技却在主流 AI 行业会议中显得格外边缘。随后，卡斯拉提议全场为基兰·布拉曼丹（Kiran Brahmandam）与厄尔·霍尔（Earle Hall）精彩的行业洞察报以热烈掌声，并正式开放现场观众问答。 --- ### [36:35](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=2195s) 问答环节：如何利用 AI 重塑玩家生命周期与流失预警 一位现场听众提问：“如何利用 AI 来重新激活（reactivate）那些已经流失或不活跃的玩家？”基兰首先指出，解决流失问题的第一步根本不在于“事后召回”，而在于“从一开始就防止他们变成不活跃状态”（make sure they don't become inactive）。在回答前，基兰特意询问了提问者是否来自博彩行业，当得到肯定答复后，基兰打趣道：“这下感觉像回到传统的博彩行业大会了！因为面对那些不了解博彩业的普遍听众时，我们总是很纠结到底该聊些什么深浅的内容。” 随后，基兰对传统赌场极其粗放的生命周期管理进行了深度剖析。大部分赌场目前依旧依赖僵化的“固定时间窗口”来给玩家打标签（如活跃、高风险、已流失或即将流失）： - 如果玩家在过去 90 天内到访过赌场，就会被系统机械地判定为“活跃玩家”； - 如果玩家在连续 365 天（或 180 天）内没有到访，才会被归为“不活跃”或“流失”状态。 这种简单粗暴的一刀切做法完全无视了玩家个人的行为习惯。基兰给出了鲜明的对比案例：假设一位客人的正常习惯本来就是“每年只来一次”，在固定周期的逻辑下，到了第 365 天他很可能会被系统认定为“不活跃”，营销部门便会投入远超其实际价值的高昂再投资成本（reinvestment）去疯狂挽回；反之，若一位“每天必到”的忠诚老客户突然连续好几天没有露面，系统由于还在等待 365 天或 180 天的硬性时间截点，根本无法察觉异常，赌场白白错失了最佳干预窗口。基兰强调，真正的 AI 赋能在于构建千人千面的生命周期模型——透彻学习每位玩家独有的行为节律与到访周期，当 AI 捕捉到行为轨迹与个人基准发生偏离的瞬间，立即介入干预，从而在客户彻底流失之前完成精准留存。 - **传统划分弊端**：传统博彩营销将所有玩家强行塞入统一的时间大桶（“huge basket or bucket”），严重忽略了个体行为节奏的异质性。 - **AI 动态干预优势**：基于时间序列与异常检测算法，AI 能够在高频玩家刚出现行为异动时第一时间告警，并在低频玩家按部就班到访时避免过度营销浪费，重塑客户生命周期的价值管理。 问答结束后，全场再次响起掌声，专题研讨圆满落幕。 ## Notable quotes > [08:12](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=492s) 这行业的好处是赌场极其赚钱，而坏处也是赌场太赚钱了 — Kiran Brahmandam > [08:12](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=492s) 这个行业哪怕逆自身运营逻辑而行，也照样日进斗金。 — Kasra Ghaharian 的导师 > [14:25](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=865s) 它是唯一一个不太懂得如何给客户合理估值的行业 — Earle Hall > [23:05](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=1385s) 这绝对是一件好事 — Earle Hall > [35:17](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=2117s) 加入自己的盐和胡椒粉 — Earle Hall > [35:53](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=2153s) 我们实际上是本届大会唯一一个与博彩相关的内容板块 — Kasra Ghaharian > [36:35](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=2195s) 从一开始就防止他们变成不活跃状态 — Kiran Brahmandam > [36:35](https://www.youtube.com/watch?v=u5WQ0fXJlKQ&t=2195s) 这下感觉像回到传统的博彩行业大会了！因为面对那些不了解博彩业的普遍听众时，我们总是很纠结到底该聊些什么深浅的内容。 — Kiran Brahmandam ## Mentioned Kasra Ghaharian, Kiran Brahmandam, Earle Hall, UNLV 国际博彩研究院 (UNLV International Gaming Institute), Aristocrat, AXES.ai, Gaming Analytics, Ai4, MGM 美高梅大酒店, Buffalo 老虎机, 《赌城风云》 (Casino), 《十一罗汉》 (Ocean's Eleven), Sphere (沉浸式球体场馆), 太阳马戏团 (Cirque du Soleil), 《游戏领域 AI 现状报告》 (State of AI in Gaming Report), KPMG, AWS, OpenAI, Anthropic, 哈里·里德国际机场 (Harry Reid International Airport), OpenTable, Uber, Desert Ford Ventures, LinkedIn ## Video description Kasra Ghaharian - Director of Research, University of Nevada, Las Vegas, International Gaming Institute Kiran Brahmandam - Managing Director, Gaming Analytics, Aristocrat Earle Hall - President & CEO, AXES.ai The gambling industry is at an inflectxion point. In just a few years, AI has shifted from an emerging novelty to operational necessity - powering everything from personalization and game development to compliance and surveillance. But as adoption accelerates, key questions arise: How are organizations deploying AI in practice? Where are the biggest challenges? How can we ensure responsible implementation? And what new AI advancements will drive future innovation? In this session, we present the results of the UNLV International Gaming Institute’s 2026 State of AI in Gaming Report. Joined by AI leaders from some of the world's premier gambling companies, we will discuss how AI is redefining the industry and helping optimize the modern casino floor. [Source](https://www.youtube.com/watch?v=u5WQ0fXJlKQ)

[8:12] Kiran Brahmandam: 这行业的好处是赌场极其赚钱，而坏处也是赌场太赚钱了 [8:12] Kasra Ghaharian 的导师: 这个行业哪怕逆自身运营逻辑而行，也照样日进斗金。 [14:25] Earle Hall: 它是唯一一个不太懂得如何给客户合理估值的行业 [23:05] Earle Hall: 这绝对是一件好事 [35:17] Earle Hall: 加入自己的盐和胡椒粉 [35:53] Kasra Ghaharian: 我们实际上是本届大会唯一一个与博彩相关的内容板块 [36:35] Kiran Brahmandam: 从一开始就防止他们变成不活跃状态 [36:35] Kiran Brahmandam: 这下感觉像回到传统的博彩行业大会了！因为面对那些不了解博彩业的普遍听众时，我们总是很纠结到底该聊些什么深浅的内容。

[0:04] Dr. Kasra Ghaharian: bursting at the seams [2:55] Earle Hall: money laundering and addiction [2:55] Kiran Brahmandam: doom scrolling [6:26] Earle Hall: the bible [8:12] Kiran Brahmandam: the industry makes money despite itself [8:12] Earle Hall: data hungry [19:44] Kiran Brahmandam: cost-cutting [32:04] Kiran Brahmandam: add their own salt and pepper

[0:04] It's really a pleasure to be here. Um, as  Mercedes introduced me, I am Dr. Kazra Gajarian. [0:09] I'm director of research at the University of  Nevada, Las Vegas International Gaming Institute, [0:14] which is literally just 10 minutes east of here.  Uh, you can see it from the strip. In fact, [0:21] um, it's really great to be back at AI4.  I think I attended my first one in 2023, [0:27] uh, when it was at the MGM. Uh it appeared to be  bursting at the seams then it's certainly bursting [0:31] at the seams now. Um I think I was on an escalator  yesterday and I got to the top of the escalator [0:37] and as we were met by a crowd of people that had  come to a standstill which was quite scary. Um I [0:42] think they've got people manning the elevators now  which is good. Um but yeah, thank you for joining [0:49] us today. I know you've got a lot of content to  choose from. Uh so thank you for spending the next [0:53] 40 minutes with us where we're going to talk about  the state of AI in gambling. Um, and I I'm biased, [0:59] but I think we have two of probably the best  people in the industry to talk about how AI I [1:05] think the two is correct. Um, two of the best  people in the industry to talk about how this [1:11] technology is impacting our sector. Um, and I'll  actually allow you guys to go ahead and introduce [1:17] yourselves. How about you give me your name,  your role, and in under 60 seconds if possible, [1:23] Earl? Um, what your company does for the gambling  sector. And how how about how about we start with [1:27] you Cohen? [clears throat] Can the folks in the  back hear me? Okay. So I'm Kiran Braandum. Uh I [1:35] am the managing director of gaming analytics at  Aristocrat. For the folks uh who may not know, [1:40] Aristocrat is a a pretty large, you know, gaming  company that makes, you know, board loads of [1:45] slot machines for the land-based casinos, online,  social gaming, and so on and so forth. Um buffalo [1:50] buffalo. [laughter] Yep. Uh and then um prior to  that just a few months before I was actually had [1:58] my own startup. It was called Gaming Analytics. We  help casinos use AI to transform their operations [2:04] and we've been you know in the business for about  8 years and then recently acquired by Aristocrat. [2:09] Yeah. So thank you Karen and hello folks. Hope  you're enjoying the air conditioning. It's a [2:17] little bit toasty outside. Um, three of us are  from Las Vegas. All three of us are pretty much [2:24] involved in regulating uh, advising the gaming  industry on all kinds of standards. And the part [2:32] that my team does is we've built a cloud cashless  and AIdriven ERP to drive landbased gaming. So [2:42] that's what I do as a day job and I'm involved in  blockchain and other different standard committee [2:47] at the same time. Great. Thank you. You both kept  to the time which is awesome. Um so I just wanted [2:55] to start by providing a little bit of a level set.  Um I'm not sure how familiar you are all with the [3:01] gambling sector. Um you know the the picture  people might have of the gambling sector is is [3:07] maybe perhaps decades old. You know something from  the movies like Casino, Oceans 11, those types of [3:14] movies. uh maybe a business that is not really  that scientific, maybe a lot of decisions being [3:20] made on intuition rather than data and science. Uh  but but the actuality is it's a highly datadriven [3:27] industry and one that has reached over $700  billion in revenue this year and [snorts] is [3:33] on track to reach over a trillion uh by 2030. Um  so you know this is a really big opportunity for [3:40] AI I think. And before we get started, I guess  um Kieran Oh. Um what's the biggest misconception [3:47] people have when you tell them what you guys do?  Kieran, do you tell people you work in gaming [3:51] and what do they usually say? Well, you know, we  get a little bit of gaming, right? Uh is that the [3:58] vertical that you are in? But um in terms of the  uh the misconceptions also, right? uh is it's not [4:05] necessarily seen as a very tech oriented industry  like based on uh you know what Kasra has uh [4:12] mentioned in terms of general image of gambling um  and then also you know people don't appreciate how [4:18] much time effort and money the industry spends on  regulating itself and complying to all the various [4:25] regulations probably more than you know some of  the big companies that you guys are you know aware [4:30] of uh in the outside world uh if you're used to  doom scrolling and all of that good stuff, right? [4:36] So the industry spends a lot of time and effort  on making sure that you know the players not only [4:40] have a great experience we do it very responsibly  and I don't think that's kind of appreciated [4:45] generally from the outside world and also the  revenue the industry generates funds you know [4:50] so many public school initiatives roads and so on  and so forth in in so many states you know right [4:57] oh anything to add so coming from Quebec Canada  originally um most people think gaming is I don't [5:07] call it gambling, I call it gaming. Most people  think gaming is money laundering and addiction, [5:14] which is so not true. We're in one of the most  highly regulated industries in the world. If you [5:21] want to work in this sector, they grab your bank  accounts. They g grab your Visa cards. They check [5:27] with all your neighbors to make sure you're an  okay person. We are in an overregulated industry. [5:34] That's the very first thing. The second thing is  people don't realize that we're in the deepest, [5:41] richest data consumer experience that exists.  A player walking on a casino floor may be doing [5:49] 10 things at the same time. They're reserving a  restaurant. They're extending their hotel. They're [5:56] going out to the sphere. They're going to suck  to so the player that's on the gaming floor is [6:04] probably one of the most retail datari data points  or avatars that you could ever imagine. And that's [6:12] the exciting thing about the industry we're in.  The limitations on what you can do to enhance that [6:19] consumer experience are absolutely limitless.  Great. Fantastic guys and great segue for my [6:26] next slide. Um so you know Kieran I memorized your  slide Kieran Earl and I you know we're we're very [6:33] used to speaking at gambling conferences so this  is quite a different audience for us. So you know [6:37] we spent a bit of time preparing and we wanted  to make sure that um you know the audience today [6:42] left knowing three key things about the gambling  sector and both Kieran and I have already touched [6:47] on those three things and here they are. So really  today we want to be talking about data safety um [6:54] and experience and and just just to get kind of a  pulse for who's in the room like so so does anyone [7:00] work in the gambling sector in the room? Oh, so  we do have quite a few. So and and who of those [7:07] um how many of you kind of build technology or  or technical leads building solutions? Okay, [7:13] great. Okay, so I think we have a nice bit of  content for everyone actually. So this is going [7:17] to be a great chat for you all. We didn't ask  who gambles, right? We did not. Who gambles? Um, [7:26] so we're going to be chatting, but we're going to  be having some slides to kind of like uh work off [7:32] of and we're going to be presenting some data from  a report uh my institute just released earlier [7:37] this year in April. Uh the state of AI in gaming.  Um these are not my words, but it's been referred [7:43] to as the bible uh for AI in the gambling sector.  my words because it's one of the best documents [7:48] I've read in all sectors combined. Well, thank  thank you very much. So, a lot of work went into [7:54] it. It's a great report. Um, we're going to be  sprinkling out some um data from that report in [8:00] today's fireside chat, but you can download it.  It's free freely accessible um at this link. Um [8:06] so, yeah, enjoy reading. It's over 100 pages long.  Um [sighs] okay, so let's start with the first [8:12] topic of the day, uh data. And um we're going to  be arguing that this might be the richest consumer [8:17] data set almost nobody um outside this room is  talking about. Um and you know it hasn't always [8:23] been that way, right? If we look kind of like  pre-200s, you know, late 1900s, uh we had a very [8:30] kind of mechanical casino floor. Um you know, the  one-armed bandits where you literally had to pull [8:34] a lever, uh put coins into the machine. Um, in the  2000s and '90s, we started to move towards Tito, [8:42] casino management systems, maybe digitizing the  casino floor so you can track players. Um, so, [8:48] so now AI has arrived uh with arguably a  data ready casino floor. Uh, but I'm going [8:56] to [snorts] stop there. Um, and Kieran, you guys  have kind of lived this arc, so you might be able [9:00] to speak more granularly about it. Um has this  transition from kind of mechanical to digitization [9:07] been seamless? Um and is the data really there? Um  even you can start. Okay. I think for me to talk [9:15] about mechanical to digital it's probably prior  my time but [laughter] uh but um I can tell you [9:22] that um you know the industry when I came in  had the same problems that any other industry [9:28] has right data silos and data is across you know  tens of systems if not hundreds of systems right [9:34] uh and then the casino operators always actually  had a challenge to bring in all this data together [9:40] and make decisions and and and by the way those  problems are not unique to to this industry it's [9:45] you know it's the same across the board. Uh  but what happened in terms of the shift in in [9:51] in trends in the last 5 to seven years is that  more and more operators started thinking about [9:57] you know how do we actually leverage this data  to act either make more money or improve their [10:02] player experience and one of the things you guys  have to know right we in a very unique industry [10:07] uh and there are good and bad with this industry  the good is casinos make a lot of money right the [10:12] bad is casinos make a lot of money so when you're  making a lot of money why should I change? Why [10:18] should I still, you know, whatever processes I've  been using for the last 10, 15, 20 years that made [10:22] us so much revenue, why should I change, right?  Uh whereas in the outside world, you kind of [10:28] squeeze for, you know, every penny, especially  on the retail side, right? This industry has a [10:32] luxury of letting technology pass by for a few  years and then catch up, you know, after that, [10:38] right? So it's it's okay to be slightly behind but  that mental shift I've seen in the last you know 5 [10:44] to seven years where more and more um the way the  casinos connect with the player is not just at the [10:51] slot machine or at the tables it's it has become  online it has become social. So your player is in [10:56] so many different channels. So the need to connect  all of this data to create a much more enhanced [11:03] experience has significantly changed in the last  four or five years and generally the industry is [11:07] very excited about see uh to see what this agentic  AI and all the future is going to bring. Yeah, [11:13] I I think a old professor of mine used to say uh  the industry makes money despite itself. But yeah, [11:20] anything to add there? So, probably the first  thing that most people don't realize about the [11:26] gaming industry is we're data hungry. Who's  on the casino floor? What are they doing? Are [11:35] they happy? Are they safe? Do they need to be  protected? Our industry is hardcoded with an [11:42] ethics of making sure that they're checking on  people to be safe and then they're trying to [11:49] make sure they're having a fun experience. You  don't go to a casino to make money. You go to [11:56] have an experience like you go to a restaurant  or anything else. If you come out with money, [12:02] if you're one out of a million people that come  out with money, well, good on you. So the thing is [12:08] is we're in a data hungry industry. So is the data  there? In 2026, we can say the data is definitely [12:18] there because even though there's silos in food  and beverage, there's silos in hotels, there [12:24] might be silos in table games, there is a way to  go find and get that data out because everybody is [12:31] hungry to know who the player is and how are they  doing. So that's one of the exciting things about [12:39] our industry. You could look in transportation and  pharmaceuticals and things like that. But the fact [12:45] that we're hungry for data, the fact that there  is so much regulation around anti-money laundering [12:51] and responsible gaming, we are genetically a data  hungering industry. Perfect. So we've kind of set [13:02] this stage where the data is. Sorry, hold on. My  slides isn't going. There we go. Okay. So, we said [13:07] the data is ready and of course, you know, we all  know by now uh so if we want to leverage AI, yeah, [13:13] it might be a bit small. Sorry about that. But, um  do the same thing. To leverage AI, we need data, [13:18] right? That's fine. Um so, we asked this in our  report. So, we surveyed over 80 companies um about [13:26] their AI adoption and their AI use cases. Uh this  was a one response per organization survey. So, [13:31] we had really high quality data. you know these  surveys were answered by CEOs, chief AI officers, [13:37] CTO's uh those kind of levels of individuals.  So the the data this uh report represents is [13:42] quite robust. Um so so we mapped use cases across  these 80 different companies. Um and what we found [13:48] was actually quite interesting. It wasn't kind  of this customer experience that uh Kieran and [13:53] have described. We we found heavy use cases and  activity uh in kind of back of house functions, [13:59] technology and security, product development and  innovation. Um we actually found less activity in [14:04] kind of customerf facing functions and then you  know the least activity was kind of in these risk [14:10] and compliance functions. Um so guys looking at  this and and what I just said is a key takeaway [14:15] in that AI seems to be being leveraged more at  the back of house rather than the front of house. [14:21] um is that what you guys are seeing and and  maybe could you explain that disparity? Yeah, [14:25] I think it depends on you know the the the the  percentage of the the the type of people that [14:32] actually answered the research paper. So if you  have asked the question to most of the vendors [14:37] right that providing software hardware to the  casino industry yes they are spending a lot of [14:42] their time on technology software development  security and so on and so forth right but if [14:47] you actually go look at how operators use the  technology right they themselves may not be [14:53] using AI the way we wanted to use to manage their  back office operations but to manage their casino [14:59] floors right they've been doing a pretty decent  job in the last four five here. So for you know [15:04] I can speak from my experiences. So uh when  you know we started the company and we started [15:09] helping casinos change the way how they make their  decisions um uh and we are about 140 casinos which [15:16] is call it 15% of the market or whatever right  so that's a pretty significant chunk in terms of [15:23] market adoption that's leveraging AI to say how  do I change my marketing campaigns what kind of [15:28] slot machines do I buy where do I place them how  do I price them how can I do real time you know [15:33] offer recommendation to the players how can I  detect fraud odd, right? So, those are all the [15:38] things that um that the vendors, sorry, the casino  operators are using products from companies like [15:44] us. But if you if you ask a question saying how  are you using AI in your back office, the answer [15:50] is probably going to be not much yet, right?  So, that's where the discrepancy comes from. So, [15:56] of course, the question the the answer has to be  regional. The questions we're getting asked in [16:03] Australia are different than Singapore because the  different cultures have different preoccupations. [16:09] The number one thing we always get asked if  we can do is predictive anti-money laundering, [16:16] which is one of the easiest things you could  ever do from an algorithmic point of view. [16:21] Second of all, jurisdictions like in Australia,  like in the UK, Singapore, they're all getting [16:27] very excited about the fact that you can use  more and more data to predict if somebody is [16:35] trending towards a dangerous experience where  they're hurting themselves by spending outside [16:42] of their window of comfort. But really, the  most thing we get asked is, who's our player? [16:50] Who's that person out on the floor? When do they  come? Why do they come? How long do they stay? Why [16:57] are they stay? Why did they just get up and leave?  So like every other retail world experience, [17:05] the number one question we get asked every day  is how do we ask these questions to the system? [17:12] Because at the end of the day, the more you know  your customer, the cheaper it costs to cater to [17:18] to their specific needs and the better they're  entertained and the safer they are. Just to bounce [17:25] off of that, uh, you know, some people have told  me this is the only industry that doesn't really [17:31] know how to value their customers. Would you  agree with that statement based on, you know, [17:36] like theory? 100%. 100% because there's there's  something we have in player theory called customer [17:44] journey and even though there's some amazing  solutions out there between the solution and a [17:53] lot of the clients there's still an education gap  of them learning what a player's journey is inside [18:00] of a session inside of a visit and over time so  I think that's one of the exciting things that [18:08] our industry is starting to embrace now is how  do I engage a long-term relationship with that [18:14] person and I think we're in the right time in the  consumer world for that because whether it's the [18:21] Starbucks app whether it's the BMW app everybody  right now from a retail experience is asking that [18:28] same question I think the exciting differentiation  in the gaming industry is the amount of data [18:36] we have to create a deeper and a more engaging  experience. Perfect. Thank you. Um, moving on, we [18:44] asked another question to our 80 or so companies.  We we kind of wanted to drill into kind of what AI [18:50] exactly is being leveraged by these companies and  uh we had various options. Um, I think the most [18:55] you know striking striking result here from our  perspective was the lack of adoption of agentic [19:01] AI. Um and you know if you've been here at the  conference there's a number of sessions on aentic [19:07] AI um and you know we did this report with KPMG so  we were able to benchmark kind of these industry [19:13] results against other industries and sectors and  we do in fact see that the gaming industry appears [19:18] to be lagging behind in terms of their adoption  of aentic systems. Uh Kieran I'll start with you. [19:24] Do do you have any idea as to why uh there might  be this slow adoption of agentic systems? Well, [19:32] you know, uh if you just talk I think the whole  word agentic AI and the and and the hype and [19:38] the value creation about that probably started  just a few months ago, maybe a year ago, right? [19:44] But leaving that aside, see the way we consume  AI in our daily lives is very different than [19:52] the way businesses consume AI, right? So if I  have an agent that goes and you know looks at [19:58] all my emails, gives me what I'm supposed to  do for the day or makes a hotel reservation, [20:02] those are all simple things. But when you come to  the casino world where it's highly regulated and [20:09] there is a very uh high need for validating and  auditing everything every decision that was taken. [20:17] These systems are not there yet to help casinos  make those complex decisions. So for example, [20:21] I want to spend $20 million in my marketing budget  this month. You know what campaign should I run? [20:27] You know, could I just have an agent go figure  out what players need to get an offer and make [20:31] it happen? And if it works, great. But if it  doesn't, you know, do we have the intelligence [20:35] to actually be able to explain what happened  and would I want to take that risk? So the scale [20:39] of the problems are extremely different in our  industry because when you make certain mistakes, [20:45] your gaming license could be cancelled. Right?  So the cost of a mistake is very high in this [20:50] industry. Very similar to the cost of a mistake  is very high in the medical industry. Right? So [20:56] that is why and and I've seen in this conference  that people have not fully solved using agents [21:02] to solve problems because they're like oh by the  way these agents are not trained on your your your [21:07] organizational data you need to have your own  context layer and by the way you have to have [21:11] all of that in a knowledge graph and have queries  go through subgraphs. So the the world is still [21:15] trying to figure out how to solve these problems.  Whereas you know the things that most of us use [21:21] for agents in our daily lives they save time but  they're not making consequential decisions right [21:27] so that's the gap I think the tech has to evolve a  little bit more um uh while the world is figuring [21:32] those problems out anybody who wants to actually  come into this industry right uh some of the [21:38] vendors that are out there just know that you know  uh in every other industry yes I can save cost [21:45] you know you can sell your solutions that way but  But in our industry, it's a little bit different. [21:49] You know, casinos make good money. And a lot of  tribal casinos, right, the native nations, their [21:54] focus is just not cost cutting because they want  to make sure they invest in their people, they [21:58] invest in the tribe members, they all have jobs.  So the standard pitch that you that you that you [22:04] give for a typical business is very different and  may not fly in the gaming industry. And accuracy, [22:09] reliability, and auditability is extremely  important. And if I ask the question, I need to [22:14] get the same answer every single time. Otherwise  your products will be canned by the regulators and [22:19] you're out of the industry. Right. So so it's  a different scale of the problem that we are [22:23] dealing with. I don't mean by scale in terms of  mass amounts of data and billions of transactions [22:28] but the cost of a mistake. Right. Yeah. It's a  highly scrutinized industry. We've got regulations [22:34] in place. So maybe there's hesitancy driving  that. I guess some um you know some data from [22:39] our survey that could maybe help uncover some more  reasons. um we kind of created this maturity index [22:46] for the industry as we called it. Um so we scored  every organization on four different pillars um [22:53] strategy, infrastructure, expertise and governance  and and kind of the headline finding was that you [23:00] know ambition is kind of outpacing everything  else. Um infrastructure expertise and governance [23:05] in particular is lagging behind strategy. Um, oh,  so you know, kind of related to what I just asked [23:11] Kieran, um, but also is this a bad thing? Do you  think the gaming industry is lagging behind other [23:18] sectors? Um, do you think they need to catch  up on things like infrastructure, expertise, [23:23] governance? Um, is this similar to other  sectors? So, my answer is going to be based [23:29] on fresh information this morning from a closed  meeting that we had with AWS. They had a they had [23:35] a really good event this morning at breakfast and  they're saying all industries combined only 7% of [23:45] agentic projects ever make it into production. 85%  make it to a pilot but only 7% actually make it [23:56] into production. Now to reinforce the point that  was just made, we are in a highly regulated and [24:06] self-regulated industry. And when you go to one  of our industry conferences and you have a casino [24:12] that wants to do an agentic project and they go  to the regulators or the state body and they say, [24:18] "We're thinking about doing this. Why do you  think about this?" And everybody just looks at [24:22] each other and goes, "We don't know." Because this  is new. We play with real money on the floor. We [24:30] play with people's livelihoods. We play with their  entertainment. So the amount that we're careful is [24:38] more than the normal than you would find in other  segments. So we're cautious as an industry. We [24:46] innovate, but we take our time to do it right. And  I think that's where you look at responsible AI. [24:54] That's where you look at ethics. We are very  in our industry. I'm on a couple of different [25:00] associations. We take our time to make sure that  we're not experimenting with our players. So, [25:09] if we say we're lagging behind other segments, I'd  absolutely say that's a good thing. Okay. Kevin, [25:18] anything to add there? No, you're good. Okay,  we can move on. I think that was good. Um, [25:22] so let's move on to our next kind of topic. We  said we're going to talk about three things. Um so [25:28] the second thing we want to talk about is safety.  Um I think this is quite timely actually. You know [25:34] there's lots of sessions at this conference about  AI safety, AI alignment. Certainly, you know, [25:40] AI is having to tackle this head on. Uh the  gambling industry, you know, we can say that it's [25:45] had to tackle this for decades since its inception  perhaps. Um, we've always been highly scrutinized [25:51] uh for things like fraud, anti-money laundering,  uh, being able to protect players from addiction [25:56] and that kind of thing. So, you know, this is  an industry that really has to prove it. Um, [26:03] so I'm really having trouble with this clicker.  Um, so in terms of AI safety in particular, um, we [26:09] actually surveyed some regulators as well as part  of our report and we asked them various different [26:14] questions. Um what was interesting is that on the  whole regulators kind of agree that there needs to [26:20] be some AI specific regulations for the gambling  sector. Um and they even think they might need to [26:25] be stricter than broader AI regulations. Um they  also don't have high confidence in the industry, [26:33] the operators, the suppliers being able to  self-regulate. Um Kieran, what do you think [26:39] of that? Do do you think that's uh reasonable uh  for the regulators to have that perspective or do [26:44] you think the industry can uh provide some good  self-regulation? Interesting question. Uh because [26:51] there's so many vendors in this industry. I can  speak for you know what Aristocrat does right. [26:55] I mean uh responsible gaming is one of their you  know core pillars and core part of their strategy. [27:01] uh and uh they spend a lot of time educating the  players on the land-based side, self-exclusions [27:07] tools, you know, early detection of a player that  is likely to become, you know, likely to have a [27:12] gambling problem. So, Aristocrat as a company, you  know, spends a lot of money and also, you know, [27:16] it funds UNLV and a bunch of universities to  actually do research on what's happening. Um I [27:23] don't have metrics on what other companies do but  generally speaking you know we've been reasonably [27:30] good in terms of self-regulating uh now if the if  the if the regulators think you know the vendors [27:36] you know are not well equipped you know I give it  to them but I I would say you know from aristocrat [27:42] perspective it's it's pretty heaven and the other  thing I would also put it back on the regulators [27:46] is that you know they also have to actually get a  lot more educated on AI and they should actually [27:51] be doing these sessions on how AI should be safely  implemented and use gaming industry as a test bed [27:58] for the rest of the world also. And so yeah,  do do you guys think there's something for AI [28:03] as a whole to learn from the the gambling sector?  You know, if there's a new game or a new product, [28:08] um there's various different gates that a company  has to get through to get that to the casino [28:14] floor. Whereas, you know, even in the keynote  this morning, Fay and and Jeffrey were talking [28:20] about how there might need to be regulation um  for for model releases, but there really isn't [28:25] any. I don't think Anthropic and Open AI can  kind of just release a model if they want to. [28:30] They don't have to get approval from anyone.  Um so, so could general AI have something to [28:36] learn from how the gaming industry works in terms  of its licensing, you know, GLI certification, [28:41] that kind of thing? Oh, do you want to maybe add  to that? So, it it's natural for us being in the [28:48] gaming industry to speak about how we're regulated  because we're way more regulated than most other [28:54] industries. Even if you look at like medtech  and the FDA certification, I would say clearly [29:03] the gaming industry has much more regulation  than that. So, can we self-regulate? Well, [29:12] in our industry, the companies that are regulated  are following the strictest guidelines already, [29:19] which means if they're implementing AI solutions,  they are already implementing AI solutions inside [29:27] of their regulations of how they treat  their players, how they build their games, [29:33] how do they play games, how do they treat their  employees. So the regulations are so strict that [29:39] in my opinion AI is just another tool like the  others as part of the regulation culture if that [29:47] makes sense. Where our gaming industry is now  trying to wrap our heads around a new subject is [29:57] where's the data? Where's it going? Who's allowed  to see the data? Who's not allowed to see it? Do [30:03] any third parties have any access to this data? So  the really cool discussion that's emerging in 2026 [30:11] is not really the AI discussion. It's the data  discussion. And that's really the core discussion [30:18] that needs to be had more than AI in our industry  because of the fact that I think we're so more [30:25] advanced than other segments when it comes to how  we regulate everything we do. Kieran, do you think [30:33] it's a stifling of innovation? Like, you know, as  I said, Open AI, Anthropic, they they seem to have [30:39] free reign. They're like not really regulated,  whereas you you you operate in a heavily regulated [30:44] industry. Do does do you feel it stifles your  ability to innovate? [clears throat] With the [30:50] realm in which I play, uh, in terms of optimizing  casino operations and all, no. But in the realm [30:57] of making games right uh obviously there are  lots of you know rules and regulations to make [31:04] sure you know we follow the responsible gaming you  know the principles on the game side yes I'm sure [31:10] you know everybody fights for you know slightly  lower regulations uh but on the system side which [31:16] is managing all the back office operations uh we  are able to you know innovate a lot more on the AI [31:23] side uh even building you know deploying some of  our you know agentic solutions um to solve some of [31:29] to to answer some of the strategic questions that  an executive may have. I don't see that much but [31:35] definitely on building the actual slot machine in  the game right now that also you know makes it a [31:42] little bit difficult for anybody to just come into  the industry because you know uh there are pros [31:46] and cons of regulations guys you know so perfect  so I went on a bit of a tangent there that was [31:52] unprepared but you guys handled it uh really well  um I want to leave some time for questions to see [31:56] if anyone has any but um we'll quickly go over our  last um topic for the day um which is experience. [32:04] Um, so really I I don't really have much to say  here, so I was just kind of going to tear it up [32:09] for you guys in terms of like what does the casino  floor or the gaming industry in general look like [32:15] 3, five, 10 years from now? You know, oh, let's  start with you. I land at Harry Reid airport. [32:22] What's my experience? So folks, this is where  it gets exciting because can you imagine you're [32:28] landing in the universe capital of entertainment  called Las Vegas. So there's absolutely an [32:37] infinity of things you can do in this town.  So where our industry is going to and the [32:44] Las Vegas tourist agencies and everybody else is  wrapping their head around how do we cater to all [32:51] of the millions of people that are coming here as  individuals. So it shouldn't be even when you land [32:58] here because if you remember the good old days  before co you get in the yellow taxi you'd get the [33:03] little booklet with the coupons and then you'd  figure out your own journey through the town. [33:09] But as soon as you decide to go to Las Vegas,  you should be assigned your own personal avatar [33:17] VIP butler. And this VIP butler can do anything  for you. This is not uh futuristic. This is what [33:26] the students at NL UNLV, our university here in  town, this is what they're talking about today. [33:33] So you can say, I've got two hours to kill. What  should I do? Oh, would you like me to book you [33:38] a VIP walkthrough of the sphere? So, the thing is  is as this town captures the data spirit of being [33:49] the center point of the customer experience, what  we'll be doing in this town over the next three [33:56] years is limitless because once the avatar becomes  the lead on your experience, there is no limit. [34:05] Just look what OpenTable has done to trying to  figure out how to reserve a restaurant or Uber has [34:12] done. Can you imagine pooling all of that inside  of one deep LLM experience of question and answer [34:20] and trying to figure out how to make sure you have  more fun that is happening right now. Like Desert [34:26] Ford Ventures has a project up and coming here  in town. This town is geared to entertainment and [34:33] it's coming to predictive LLMs real real soon.  Great. Great. So, so that was like kind of the [34:40] destination perspective I'd say. Kieran, maybe  could you speak to like the you know we walk [34:45] into the casino maybe what what do you envision?  Maybe it could be from a business or customer. I [34:50] think you know I I focus a little bit more on  you know experience from the operations side [34:55] right on the back office side because we focus  so much on the players but not actually people [35:00] who make decisions on how the players experience  should be. Um so that's where generally you know [35:05] my focus is and what we envision you know how  casino operations run in the future and for that [35:11] matter any business hopefully right is you know an  executive saying can I get an extra 3% revenue you [35:17] know this month and then we have AI and all these  agentic solutions you know figuring out recipes [35:23] that we can actually recommend to that executive  and the executive can actually add their own salt [35:28] and pepper to each of these recipes adjust it and  push a button and then through technology ology [35:33] you know offers are sent various integrations  are happening a text message comes to the player [35:38] hotel reservations are made so that is where you  know I think you know the world is going to go in [35:44] our industry where operators are going to define  business goals and software is going to execute [35:49] it and that kind of tech probably is pretty  relevant to the rest of the world also right [35:53] that completely changes how you run your business  great no thank you guys um so we finished with a [35:59] couple of minutes so I think we do have time for  some questions I'll just put our um LinkedIn links [36:04] up there. So, if you want to connect with any  of us, you certainly can and I'm sure we'll be [36:08] happy to talk to you afterwards. Uh but thank you  all for joining today. Um you know, we are in the [36:13] entertainment capital of the world, maybe the  G definitely the gaming and gambling capital of [36:18] the world, yet we're heavily underrepresented at  this conference. I think we are actually the only [36:22] gambling related content at the conference. So,  thank you all for joining today and uh if you'd [36:27] like to give Kieran and L a round of applause  and we'll take any questions you might have. [36:35] So the question is how do you leverage AI to  reactivate you know an inactive player? I think [36:41] the first step is trying to make sure they don't  become inactive right. Uh so so one of the are you [36:47] from the industry sir? Okay. Uh so now this feels  like I'm a traditional gaming conference because [36:53] we're always struggling. What do we talk to this  audience who are generally not involved with [36:57] the industry? So typically what happens in most  casinos is the way you define whether a player is [37:02] active or high-risk or churned or past you is rest  of the guys you know this is how it is based on [37:10] fixed time periods meaning if I have a player that  comes to my casino in the last 90 days for example [37:16] they're active if I have a player that I don't  see them for 365 days or whatever these fixed time [37:21] periods are right so the way the life cycle is  classified in this industry is defined by standard [37:26] fixed time periods Right? So, we're taking a  bunch of players and ignoring their individual [37:31] patterns and putting them in this huge basket  or bucket and then reacting to them, you know, [37:35] and and sending them off or securely. So, for  example, if your definition of inactive is players [37:40] who don't show up in 365 days or 180 days, what  if I'm a player that only comes every year? So, [37:46] technically, as far as I'm concerned, I'm not  inactive, right? I'm still an active player. [37:50] Today the way decisions are made in the marketing  side is I get reinvested a lot more than I should. [37:57] Whereas if a player that comes every day stops  coming, you're still waiting 365 days to figure [38:01] out that they're inactive. Right? So the fixing  the life cycle by understanding every player's, [38:07] you know, patterns and then leveraging AI  to figure out a change in behavior and and [38:13] reacting at that particular point is the way you  fix it, right? So okay. Uh thank you again guys. [38:19] Appreciate it. Thank you. All right, let's give  the panel another round of applause. Thank you.