The Infinite Loot Box
Why AI-native products need game designers
When generative AI products emerged, they promised endless creation and productivity. Perhaps it was just my *slight* addiction to dopamine fuelled activities, But rather than becoming the next Shakespeare of Picasso, I found myself trapped in an addictive loop of prompt-and-generate. Whether crafting images, text, video, or sound, the anticipation of each potential result kept my finger glued to the 'generate' button. I wasn't creating; I was gambling. AI had become an infinite loot box.
The Prompt Casino
In video games, a loot box offers random rewards, often purchased or earned through gameplay. This mechanism mirrors slot machines, tapping into our love for chance and the joy in the reveal. AI generation follows a surprisingly similar pattern: Each AI prompt becomes a digital pull of the lever. Input your text, click generate, and wait. Will the AI produce a masterpiece? A hilarious mishap? Something unexpectedly inspiring? The unpredictability of outcomes creates a powerful intermittent reinforcement schedule – the same principle that makes slot machines so addictive.
This psychological hook explains why users continue engaging with AI tools even when not every interaction is satisfactory. The occasional 'wins' – those brilliant, unexpected outputs – become powerful motivators. We're driven by the possibility that the next attempt might yield gold, transforming what could be a purely productive tool into something game-like. Even the business model of many AI tools, charging per generation, echoes coin-operated arcades. Each 'play' costs something, but the potential reward keeps users investing. This economy transforms productivity into a game of chance and turns creative work into a coin-op adventure.
The result? A shift from purposeful creation to playful engagement. The suspense, unpredictability, and occasional brilliance of AI outputs form an irresistible loop. But why view AI through a gaming lens at all? Because it reveals a crucial insight: for many users today, AI has become more than just a tool. It's an engaging, sometimes addictive, loot box – drawing more play than purpose.
Fun is an elusive concept.
When we think of repetitive tasks, our minds often jump to tedious chores. Yet, in video games—a medium centred around entertainment—something curious happens. Completing laborious tasks, seemingly antithetical to fun, is often a fundamental part of gameplay.
This phenomenon, known as 'grinding' involves repeated actions—whether pressing a controller button or clicking 'generate'—that transform mundane tasks into engaging gameplay loops. Players willingly dedicate countless hours to activities that, described out of context, sound pretty dull: watering crops, mining for resources or lighting an entire continent's worth of fires in RuneScape.
So why do players find amusement in this monotony?
Game designers have long incorporated motivational cues, carefully balancing effort and reward to create a sense of achievement that feels earned yet attainable. Progress bars, experience points, and loot systems are crafted to tap into our innate desire for growth and acquisition.
Infinite Content Doesn’t Mean Infinite Fun
While the 'grinding' mechanic can be engaging, fun is never endless. Exposure to the same gameplay mechanics can lead to habituation—where users become desensitised to once-thrilling outputs. Over time, engagement may wane as the novelty of generation fades and the process becomes predictable. To combat this, designers must motivate players to keep going.
In game design, there are two forms of motivation for the player:
Extrinsic—the drive to accomplish goals
Intrinsic—where the activity itself is rewarding.
Grind-heavy games like Farmville or Cookie Clicker exemplify extrinsic motivation, engaging players with promises of rewards like new items or milestones. Conversely, the intrinsic joy comes from activities like side-flipping in Super Mario or solving puzzles in The Witness, where the process itself is the reward.
By mixing intrinsic and extrinsic motivators, game designers create a more balanced and engaging experience. This approach keeps players invested for longer periods, offering both the satisfaction of achievement and the joy of the experience itself. (Think filling in your notepad with drawings of various flora and fauna in Red Dead Redemption II, before going back to slaughter another town’s-worth of cowboys).
Current AI consumer products primarily cater to extrinsic motivation, with users driven by specific outcomes like creating the perfect image or problem-solving. Like grind-heavy games, they risk losing user interest as novelty fades. While unpredictable outputs can act as 'loot boxes', maintaining engagement through surprise, this isn't sustainable long-term. The difficulty lies in creating AI experiences with engaging core mechanics that extend beyond the initial excitement of generation. Truly sticky AI consumer products need to provide lasting value, balancing unpredictability with meaningful interaction, even if that mechanic is a text box.
But is it all just hype?
Owing to recent advancements in gen AI models and significant venture funding1 products like ChatGPT and Midjourney have become widely available allowing the creation of images, videos, and sound from simple text prompts. Type a few words, wait a few seconds and like magic, instant art, stories, or code materialise at your —often mutated—fingertips.
Many remain sceptical and wonder if generative AI will follow the path of VR, the Metaverse, or Web3, potentially becoming another bubble. These previous tech trends often relied on unproven concepts or repackaged existing technologies:
VR bet on consumers embracing an uncomfortable and isolating form factor (headsets).
The metaverse attempted to repackage game engines as the future UX of work and socialisation, but its empty virtual worlds proved more alienating than innovative.
Cryptocurrencies promised decentralisation of the financial system but delivered a volatile playground for speculation.
But generative AI enhances rather than replaces existing processes. It excels at recreating familiar outputs—emails, images, code—often with uncanny accuracy. It is ubiquitous; AI-generated content can frequently pass as human-made, quietly permeating vast swaths of the internet, and scarily according to one report, by 2026, a staggering 90% of online content could be AI-generated.
Partners in Creation
There's a reason classical art lacks selfies. Photography didn't just change the methods of creation; it transformed our entire mode of expression. AI will be implemented to quicken workflows - much like how graphic designers are now far more likely to own an Adobe subscription than a letterpress. But beyond the replacement of labour or human-exclusive creativity, AI is also in “the process of transforming machines —which until now, have been our tools—into our partners”2.
Unlike VR headsets gathering dust or crypto wallets lying dormant, AI applications are seeing sustained, growing engagement. A recent Deloitte survey found 79% of businesses are already using generative AI, with 58% expecting it to transform their industry within the next three years.
In the consumer space, ChatGPT's growth has been unprecedented. It became the fastest-growing consumer application in history, reaching 100 million monthly active users just two months after launch. To put this into perspective:
It took TikTok about nine months after its global launch to reach 100 million users
Instagram took 2.5 years to reach the same milestone
Twitter needed over 5 years to hit 100 million monthly active users
However, this explosive growth isn't without challenges. The computational demands of large language models translate to significant operational costs. OpenAI reportedly spends several million dollars daily to run ChatGPT, raising questions about the long-term sustainability AI tools. This financial burden has led to the absorption of promising AI startups by tech giants, as seen with Microsoft's $13 billion investment in OpenAI and $650 million acquisition of Inflection.
Yet, even as individual AI companies face financial pressures, people's habits are changing: AI is evolving into a creative collaborator and becoming something more personal - a digital companion. This shift in user behaviour suggests that AI's impact will persist, regardless of which companies ultimately dominate the market.
Playful Software
Character.ai, a platform for creating and interacting with AI personas, tops 200 million visits per month (that’s Candy Crush levels!), with users spending an average of 29 minutes per visit—a figure that eclipses ChatGPT by 300% and they are not just using it to write their homework. The growing popularity of chatbots isn't derived from promises of productivity increase or tangible outputs, but on the intrinsic enjoyment of communication.
This appeal is hardly surprising with a user base of mainly 18 to 24 year olds, most of whom identify as gamers. The enjoyment derived from communicating with these characters—essentially NPCs (non-player characters) —blurs the lines between utility and play, productivity and leisure, in unprecedented ways.
While software has long incorporated playful elements alongside its function, AI takes this a step further. It might be quicker to google a query but it is more fun to ask someone their opinion. Our tendency to anthropomorphize dynamic objects finds new expression in AI, which embodies our language in profound ways. This linguistic mimicry creates a unique form of social interaction, one that feels surprisingly human.
The popularity of chatbots is particularly intriguing because it's an activity with no tangible output. Unlike AI tools for image generation or coding assistance, chatbot interactions produce no practical results beyond the conversation itself. While users find value in these virtual companionships, the current state of AI interaction is still rudimentary. The 'gameplay' is often solo and private, limited to text-based exchanges. Many users express frustration with the bots' lack of sophistication or consistency.
However, to maintain long-term engagement, AI tools must offer more than mere playfulness. The interaction process itself—the core mechanic—needs to be inherently fun and rewarding. This is a lesson well-learned from gaming: the most enduring games are those where the fundamental gameplay loop is enjoyable, regardless of external rewards.
A toy that can play back.
Today's 'generate and hope' AI experience—more Colossal Cave than GTA VI—is still in its infancy. Bottlenecked by cloud latency and the humongous computational demands of large models, these tools can't even match the real-time responsiveness of Pong. Yet, surprisingly, engagement metrics in chatbots reveal their possibility to tap into our primal drives for discovery, socialisation, and reward.
The future of AI, however, extends far beyond these current limitations. Defining the UX of AI-Native products requires looking beyond efficiency. Just as video games evolved from primitive text adventures to immersive, open-world experiences today’s AI-native products are in their most primitive form. By understanding the psychology of gaming, AI could evolve from a slot-machine content generator into a dynamic partner—a toy that can play back.
According to CB Insights' latest report, AI accounted for 35% of global funding in Q2 2024 - a record high - pulling in $23.2 billion, and a 59% jump from the previous quarter.
'The Age of AI' by Kissinger et al., page 20.