Skip to main content

Square Enix’s free AI game isn’t selling me on the tech’s power

At this year’s Game Developers Conference, AI made headlines. Several talks revolved around the evolving tech, with developers discussing how it could be integrated into games. That was a controversial topic during the show, but one presentation seemed to wow attendees. Square Enix hosted a 30-minute live demo of an AI mystery game that showed how the tech could be a force for good, giving players more control over what they say and having NPCs that respond to prompts accordingly.

SQUARE ENIX AI Tech Preview: THE PORTOPIA SERIAL MURDER CASE【English】

One month later, Square released that demo to the public for free. If you go to Steam right now, you can download The Portopia Serial Murder Case and try it for yourself. In fact, a lot of PC players already have … and the response hasn’t been too kind. As of the time of writing, the tech demo has a “Very Negative” rating with over 250 user reviews. While some of those could be reactionary review bombs from detractors of the tech, many appear to be legitimate criticisms calling out an incredibly limited experience.

Recommended Videos

“There might be some exciting tech behind the curtain but to end-users, it really just feels like a language engine from the late 90s,” one review reads. “It only responds to hyper-specific queries and doesn’t seem even half as flexible as they’re pretending it is.”

They’re not wrong. The educational project isn’t exactly a strong first look for the potential marriage of gaming and AI. The Portopia Serial Murder Case is an odd tech demo that seems to make the problems it aims to solve much worse. Despite that, I’m not ready to write off Square Enix’s efforts entirely based on the experiment, as I can see some areas where the tech could be useful in the long run.

Unnatural language processing

The Portopia Serial Murder Case acts as a modern remake of an old Japanese visual novel first released in 1983. The intent here seems clear: show how far video game tech has come by contrasting a restrictive, old adventure game with a new version supported by AI. In a manual bundled with the download, Square Enix gives some context as to what the project is trying to show off.

“At the time of the game’s original release, most adventure games were played using a ‘command input’ system, where the player was asked to type in text to decide the actions of their character,” it reads. “Free text input systems like these allowed players to feel a great deal of freedom. However, they did come with one common source of frustration: players knowing what action they wanted to perform but being unable to do so because they could not find the right wording.”

The introduction goes on to explain that those restrictions came down to the title’s Natural Language Processing (NLP), which was limited at the time. The new project acts as a showcase of how much that’s changed in 40 years as Square Enix shows off the improved NLP it’s currently working with. Unfortunately, the tech demo doesn’t currently impress in the way it’s meant to.

A player converses with an NPC in The Portopia Serial Murder Case.
Image used with permission by copyright holder

When I open the game, I’m thrown into a basic visual novel. I’m a detective investigating various locations across Japan trying to solve a murder by finding clues and interrogating suspects. My assumption going into it was that I’d be able to type just about anything into the text box and get some kind of relevant response. I immediately learned that wasn’t the case. Rather than functioning like ChatGPT, it feels like I’m trying to hold a conversation with Siri. Anything the game deems random gets me some boilerplate line to the tune of “Let’s not talk about that right now.” Instead, I need to figure out a pretty specific set of commands to make anything happen. If I want to ask strangers on a busy street for information, I need to write “ask around.” If I try to go off script, I won’t get the desired response. I experience a similar issue in interrogations, where forgetting to include a “the” in a command can leave the game confused.

After 30 minutes, I’d barely gotten anywhere. Nothing I was typing ended in an answer and I mostly found myself repeating the same few phrases over and over to move the story along. I’d begin to progress the more I picked up on recognized phrase templates like “look at …” but I became skeptical that Square Enix has actually solved the problem it set out to fix. My investigation felt every bit as unnatural and frustrating as classic text adventure games.

Part of the confusion I experienced seemingly comes down to a big tech omission. Square Enix noted that it had to remove the demo’s Natural Language Generation tool for now, citing the current risk of “unethical replies.” That aspect is what made the demo such an impressive showcase at GDC, so its absence here is a little confusing. What exactly are users seeing here that wasn’t already shown in games like AI Dungeon years ago?

Speech-to-text

Things would get a little more interesting when I’d enable voice commands in the main menu. With the tap of a button, I could speak prompts into my microphone and see them instantly translated into text and responded to accordingly. The mystery began to move a little more naturally at that point, as I hopped around town picking up shining clues and calling phone numbers by reading them out loud. It’s undoubtedly impressive to see spoken words clearly recognized in the game, going further than Dead Island 2’s recent Alexa integration.

Image used with permission by copyright holder

Still, that part of the demo is frustrating in its own right. Certain names would get mangled by the text-to-speech tool when I tried to speak them, forcing me to say them again and again until the computer wrote them outright. It flashed me back to the days of playing Brain Age on my Nintendo DS and struggling to get the word “blue” to register.

Though it’s positioned as an educational tool and a piece of software training, I’m not exactly sure what I’m supposed to take away from The Portopia Serial Murder Case. Simplistic language commands and repetitive, canned responses from NPCs don’t really make a case for how Square Enix’s Natural Language Processing or Understanding models can be used to improve a game like this. It’s not even clear how AI is being leveraged here — that much is clear from the Steam user reviews confused about how to interact with the game successfully.

Even so, it’s nice to see Square Enix being so transparent with its tech experiments and giving anyone the ability to play with its work-in-progress tools. This implementation is far from impressive, but I’m curious about what the publisher is attempting to solve. I’d love to play a mystery game where I can say anything I want into a microphone and have NPCs not only understand it, but respond in a relevant way (so long as scripted lines are left to human writers). That’s clearly not what anyone’s experiencing here, but hopefully, this is a necessary first step toward achieving that dream.

Topics
Giovanni Colantonio
As Digital Trends' Senior Gaming Editor, Giovanni Colantonio oversees all things video games at Digital Trends. As a veteran…
This excellent game about waiting is the anti-WarioWare
A boy waits at an airport in While Waiting.

How much of your life is spent waiting? Think about it. Waiting at the DMV, waiting for a train, waiting for an important phone call, waiting for your perpetually late friend to show up, waiting to grow up. Life is a long endurance test of patience.

That’s the simple thesis of While Waiting, a new puzzle game that’s deceptively emotional. On paper, it sounds like a one-note gag. It features 100 bite-sized levels, each tossing players into a familiar life experience that requires some form of waiting. It’s like the anti-WarioWare, swapping five second microgames for drawn out ones that require virtually no reaction time. Don’t let that light premise take your guard down, though: While Waiting is an ingenious bit of video game storytelling that’ll stick with you so long as you’ve got the patience to see it through.

Read more
One of 2024’s most overlooked games just came to consoles. Don’t miss it twice
A young girl in a white dress looks back. She carries a bag and a bow on her back while holding an orange leaf in her hand.

We are currently in something of a Metroidvania boom. With the genre reaching peak saturation, new releases -- of which there are many -- are forced to expand upon the core formula fans know and love in hopes of standing out. Take the thrilling Ender Magnolia: Bloom in the Mist as an example, which imbues the Metroidvania with RPG systems and a focus on player builds. Momodora: Moonlit Farewell, on the other hand, believes that simple is better.

Originally released in January 2024, Moonlit Farewell is the fifth and final entry in developer Bombservice’s Momodora series. While it might not have any ostentatious additions to the basic Metroidvania formula, Moonlit Farewell’s devotion to a sleek rendition of the genre’s core pillars make it a refreshing adventure. The game went under the radar in 2024 but now, with its release on consoles, Metroidvania fans have the perfect way to tide yourself over until Hollow Knight: Silksong.

Read more
Trading digital Pokémon TCG Pocket cards is a risky but lucrative business
Trade cards in Pokémon TCG Pocket.

After Pokémon TCG Pocket released its controversial trading feature, an odd sort of "black market" has arisen around some of the rarer cards. Some players have started to sell the digital cards on eBay, charging anywhere from $5 to $10 per card. Yes, it's a definite violation of the games' terms and conditions, but the real-money trades are difficult to prove and present little financial risk to the seller  — but don't think it's an easy to line your pockets.

The initial release of the trading mechanic was met with backlash from players. Developer Creatures, Inc promised changes to the way the system works, but provided few details regarding how or when those changes would take place — and while there are restrictions on how trading works, those restrictions haven't stopped players from finding workarounds.

Read more