Deep reads #6: Artificial life in a natural world

It’s been a while since the last one of these, hasn’t it? It takes a long time to put these deep read posts together, but I always feel good by the end. This time, I dive into artificial intelligence, a field I have a lot of interest in but absolutely no technical knowledge about beyond the most basic level. For that reason, I’ve tried to avoid getting into those technical areas I don’t understand well, sticking to the more philosophical aspects that I can actually sort of write about. If you know more about the subject and can bring your own perspective to the comments section, I’d welcome that.

Also, some story spoilers for Time of Eve, and very very general ending spoilers for the film Ex Machina just in case you plan to watch these and want to go in blind, which is always best in my opinion. Just being safe as usual. And now on to the business.

Sometime in the future, society has started to integrate realistic human-looking androids into everyday life. Rikuo, a high school student, relies on his family’s household android to make his coffee and breakfast in place of his seemingly always absent parents.

One day, Rikuo checks on the movements of this android and discovers that she’s been visiting a mysterious location on a regular basis, a place that he never told her to visit. After letting his friend and classmate Masaki know about it, he decides to investigate by going there himself. And so he finds Time of Eve, a café with a special rule: no discrimination between humans and androids allowed.

Time of Eve is a six-episode original anime series aired online in 2008, sometimes listed under its Japanese name Eve no Jikan. It was on my list to watch for a long time until I finally got to it last year. And while I enjoyed it, the series also raised some questions, or maybe reminded me of questions I’d already been asking myself — questions way too big for my own puny mind about the future of humanity.

Most of the action in Time of Eve takes place in the café it’s named after. Rikuo and Masaki don’t fit in very well at first, though. The lone proprietor Nagi is welcoming and friendly, but she also demands that they stick to the house rule: no discrimination between human and android patrons. This even includes asking whether a patron is human or not, leading Rikuo and Masaki to look around and speculate about all the café’s customers.

But why would this even be an issue? As Masaki explains to Rikuo, Time of Eve operates within a gray area of the law. In response to the creation of humanoid robots so realistic that they’re passing the Turing test left and right, legislators have passed laws that require they use holographic halo-like rings to differentiate them from humans. At the café, nobody has a ring, but Rikuo knows his family’s household android has been here, and considering the house rule, it’s safe to assume that at least some of the patrons are androids with their rings turned off in violation of this law.

Further complicating matters is the fact that all the café’s customers seem human enough from the way they act. When Rikuo and Masaki meet Akiko, a chatty, excitable girl, they assume she’s a human like them.

The next day, while Masaki is teasing Rikuo about his wanting to see her at the café again, Akiko shows up at their school — not as a student, but as an android to deliver something to her owner there, now with the holographic ring over her head. The pair are shocked, and the effect is made all the stranger when she doesn’t acknowledge them there but is just as friendly as before when they return to the café later.

This strangeness ends up hitting Rikuo at home when he realizes that his family’s android — after a couple of episodes finally referred to by a name, Sammy — went to Time of Eve because she wanted to make a better cup of coffee for him and his family. Rikuo first loses it, demanding to know why she was taking her own initiative without any orders. Soon enough, however, Rikuo starts to accept the situation, and we can see him thinking of Sammy as more human-like. This invites mockery from both his older sister and his friend Masaki, who say he’s starting to sound like an “android-holic”, or someone who relies too much on androids in place of fellow humans.

This fear of androids isn’t totally unjustified. Time of Eve presents a world in which these humanoid beings, far more skilled than humans in technical ability, are taking jobs, not just as household servants and couriers but also as teachers and musicians. Rikuo has already been feeling the effects of this change — it’s revealed that he gave up playing the piano because android players were starting to overtake human ones. This is a change that hits Rikuo all the harder because being a pianist was a dream of his before, one that he clearly felt was taken away from him.

Another social change, one potentially disastrous for birth rates, is the new phenomenon of human-android sexual relations. Android-holics are even referred to as living with androids in romantic relationships. These people are somewhat ostracized and are heard being criticized and mocked. However, it’s still enough of a problem that an “Ethics Committee” headed by Masaki’s father works to keep human-android relationships in line, even running ads discouraging people from seeking out partnerships with androids, as human as they might seem on the surface.

All this boils down to a question that works in the sci-fi genre have been asking for a long time now: if an android is created that acts like a human and seems to have thoughts and feelings like we do, is it any different from a human in a meaningful way? Every year, with the development of more advanced robotics, augmented and virtual reality, and AI technologies, this question comes closer to leaving the realm of fiction and entering that of reality. How will people and their governments around the world react if or when AI starts to be integrated into society itself, even into the roles traditionally played by one’s relatives and partners?

I already wrote a bit about this theme in my extended look at Planetarian, a visual novel that’s largely about the relationship between a human and an android. But that story took place in the post-apocalypse. There’s no real concern about society in that world, where civilization has already been destroyed. Looking back, the contrast with Ex Machina might have been slightly off for that reason, though I still basically stand by everything I wrote then. However, I do think Time of Eve makes for a more effective contrast because it deals with some of the same questions Ex Machina did about the social implications of the human-android relationship, but again in a very different way.

I already wrote about all the faults I found with the treatment of this relationship in Ex Machina; you can find all that in the link above. But to put it briefly, director/writer Alex Garland seems to have assumed that humans and androids can never understand or empathize with each other. At least that’s the idea I felt Garland was communicating through the ending of Ex Machina.

Time of Eve, like Planetarian, doesn’t make that assumption. In fact, I’d say the central relationship between Rikuo and Sammy changes throughout the series because Rikuo realizes from his time at the café that they can understand and empathize with each other. The fact that Sammy is an android doesn’t seem to matter by the end; Rikuo accepts that she, Akiko, and the other androids around them may as well basically be treated as fellow humans instead of mere pieces of machinery.

These deeper issues surrounding human-AI relations are still some years off, since we’re still not close to creating a convincingly human android or AI for that matter — certainly not if Sophia is the best we can do at the moment. For that reason, Time of Eve still comes off very much as science fiction to me. Unless some of the wilder conspiracy theories I’ve heard are true, we don’t have realistic-looking human-styled androids walking among us.

However, the AI musician aspect of Time of Eve isn’t quite as far-fetched now as it might have seemed 13 years ago when it was aired, because AI has actually begun moving into — or intruding upon, depending on your perspective — artistic areas that were previously thought to be purely “natural”, purely human. In the last few years, AI tools to generate images, text, and sound files have become available to the general public. I am absolutely not an expert when it comes to the technology behind these tools, but my understanding is that consumer-level AI tools can roughly imitate human-created media by using pattern recognition.

Some of these tools are pretty damn impressive. Some time ago I came across a site featuring AI-generated paintings for sale, each piece created through a process described here. Again, I don’t quite understand the specifics behind how this works, but it seems like these pieces are generated when the AI analyzes human-created art and produces something original based on a particular style.

The AI comes up with some interesting-looking stuff as well. Here’s one example I like. Quite an abstract piece as you might expect, but the AI can also produce human figures and other subjects in more classical or traditional styles.

Visual art isn’t the only place AI has dabbled either. AI-produced music has made impressive strides, putting together songs that sound like something that might have come from a human composer if you didn’t know the difference. The above piece is a pretty basic sort of instrumental rock song, something that you might expect out of a studio that produces background or soundtrack music, but the AI does follow that formula well enough to create something coherent.

The same is even true of writing. This obviously hits home closest for me, since I’m a writer. An amateurish writer to be sure, but I still take pride in my thoroughly unprofessional work full of f-words and mediocre grammar. However, I can’t ignore the fact that AI is edging in on my territory. Predictive writing AI programs like AI Dungeon and NovelAI1 are designed to build stories based on the user’s prompts. Older programs produced pretty obvious nonsense, sometimes ending with an entertainingly bad result — see the AI-written Harry Potter and the Portrait of What Looked Like a Large Pile of Ash for an example of such material. But the newest technology is again pretty impressive, producing text that’s at least coherent most of the time.

The use of emerging technology for the purposes of art and entertainment is nothing new. You could argue that this process extends back thousands of years, through the creation of new musical instruments and drawing/painting tools. In that sense, even a modern innovation like Vocaloid is just one part of that long trend. For all the concern over synthesized singers replacing human ones, Hatsune Miku and her friends are essentially just new types of instruments, only with avatars and some fan-created backstory and personality attached. The songs are still composed by humans; they’re only artificial in the sense that they use synthetic as opposed to acoustic instruments.

Miku is basically a cute anime girl vocal synthesizer you can dress up. The best musical instrument since the piano, and maybe even better, because you damn well can’t put a piano in a cheerleader outfit or a swimsuit.

In the same sense, the trend towards VTubers in place of “real-life” streamers shouldn’t be a concern for people worried about the replacement of humans with AI. Funny enough, the original VTuber Kizuna Ai played on this theme, her character being an advanced AI learning about the human world. However, the only difference between a “real” human streamer and a VTuber is the use of an avatar. The fascination with VTubers might be more a part of an escapist trend, adding an element of fantasy to streaming with its cute angels, demons, and fox/dog/shark girls.2

Even so, between the increased use of synthetic instruments and tools and emerging AI art generation technologies, it’s not hard to imagine a future in which AI can put out work that resembles human-created art closely enough that it turns from a novelty to a viable, cost-effective alternative. This may be especially true of formulaic art created for mass consumption, the sort you hear and see and don’t think too much about. And I’d say it’s already somewhat true of the more abstract-looking pieces you can find on various AI-generated illustration sites — the sort that I could imagine hanging in an office hallway or hotel lobby somewhere, a piece that might just be vaguely noticed and passed by.

There’s an obvious objection to all this: that the works generated by AI lack meaning. There’s no intent behind them. It’s true that the general form of an AI-generated work might be determined by humans, who set the parameters for the program: what style to follow, what colors or tones to use, and essentially what sorts of human art it should imitate when generating something original. But the end result is something that can’t connect with an audience on an emotional level, or at least not intentionally. We humans are great at finding patterns when we want to find them, seeing shapes in clouds, hearing hidden messages in music played backwards. On that level, it might be possible to read some kind of meaning into a piece of AI-generated art, but that reading says nothing about the art itself and everything about its audience.

Shocking news: people who think rock is inspired by Satan hear Satanist messages in rock albums played backwards! I don’t need more proof than that.

To me, this lack of intent behind these artificial pieces of art makes them feel empty. Not that I hate or even dislike them — I find some of them really interesting, but only on a technical level. And some of that interest comes from seeing how these AI-generated works differ from human ones.

I think the lack of human-like thinking and intent is most obvious when an AI tries its hand at realistic-looking human figures; the ones I’ve seen have come out close but somewhat off and wrong, especially in their faces. Not in the way a human unskilled at drawing would mess them up, either — there’s a kind of technical “skill” in the AI work if you want to call it that, but details in the figure make it clear that the AI isn’t “thinking” about what it’s drawing in the same way a human would. See Edmond de Belamy,3 an AI-generated portrait of a fictional French nobleman, and how the face is smudged. Similar paintings that try for more detail seem to do a little worse, misplacing eyes and noses in curious ways and, for me, planting themselves firmly in that infamous Uncanny Valley.

Of course, there’s a lot of argument to be had over how much the intent of the artist should be taken into account when examining art. I take what I feel to be a pretty balanced view: that both how an artistic work is meant to be perceived and how it’s actually perceived are important to understanding it. When art is put out to public view, the public takes their own kind of ownership of it in the sense that they get to interpret it for themselves. But the artist’s intent still matters. Some people may feel differently, but if there is no intent behind the art, I can’t connect with it in the same way I could with a human-created piece.

But what if the art in question is so convincing and feels so meaningful that you can’t tell the difference? At that point, does the divide between the artificial and the organic even matter? This comes back to one of the central questions asked in Time of Eve. By the end of the series, Rikuo answers this question for himself by returning to the piano and playing for the café’s audience. By returning to the music he’d previously rejected because he felt it had been invaded by androids, he accepts them.

It’s clear enough that the androids in Time of Eve are essentially human in this sense. They’re completely differently when we see them in the outside world — Sammy and Akiko both act in a sort of robotic “just carrying out commands” way while in sight of humans, as if they’d get in trouble if they acted otherwise. When they’re in the café, by contrast, they act much more naturally, as if they’re letting out their breath after holding it in for a long time. It seems that all they want is to be spoken to as equals, as though they’re humans as well; the fact that they’re synthetic and we’re organic doesn’t make a difference.4

That’s the key to that central question in Time of Eve. Its androids are self-aware and have that intent and even emotion behind their actions. I think if a real-world AI can express that intent through the creation of original art not just based on analyzing scraps of existing human-created work, that would be a sign of AI so self-aware that it might essentially be considered human in the same way.5

Of course, as far as we know, we’re nowhere near that point yet. Any AI out there that the general public knows about (leaving a gap there for any possible ultra-secret experiments in progress) still thinks like AI. When I’m out driving and I have Google Maps guiding me, it still tells me to take a left turn by swinging through five lanes of busy traffic over a few hundred feet. That direction might make sense to an AI, but any human who’s ever been in a car will understand why it’s actually a terrible direction to give.

Maybe that’s the real test: when the AI understands what I’m going through when I’m driving my car in rush hour traffic and empathizes with my experience. At least enough to not suggest such a suicidal route.

Hey Google, I get that this is technically the fastest path to my destination by one and a half minutes but maybe consider my fucking blood pressure too. (Source: B137 – Own work, CC BY-SA 4.0.)

As for Time of Eve, there is one criticism I can make: that it might be a little too optimistic, especially for the reason that it doesn’t really address the whole “humans losing their jobs to more skilled androids” problem beyond just acknowledging it. It is absolutely a problem, in some sense one we’ve been facing for centuries now with automation of work starting in agriculture and leading up to the development of advanced AI today. It’s not a problem we can’t solve, but it is one that will probably cause a lot of social strife before that point.

Then again, this series provides a nice counterpoint to all the overly pessimistic science fiction we have today, the sort that’s practically anti-scientific development. Again, I’m definitely biased on this subject, but the Luddite approach to this problem is absolutely the wrong one. We shouldn’t try to limit our development out of fear of what might happen as a result.

Time of Eve doesn’t imply that everything will be sunshine and rainbows in the future. But it does deliver a more hopeful message than we usually see out of Hollywood these days. As much of a pessimist as I am generally, I can really appreciate that, and I’d say it’s absolutely worth watching even if you end up coming to a different conclusion. I, for one, welcome our new android friends, and I sincerely hope they don’t become our android overlords instead. 𒀭

 

1 AI Dungeon and the AI writing programs that gained popularity afterward make for another potential deep read rabbit hole subject. AI Dungeon was previously the premier AI story creation tool, but developer Latitude placed sexual and other mature content control filters on the program leading to suspensions, bans, and an exodus of users to alternative services.

I’ve messed around with both AI Dungeon and the much newer NovelAI, and while they’re interesting (well, AI Dungeon was interesting before it was utterly fucked by its own developer — the filter was supposedly meant to prevent certain types of extreme/gray-area material from being written, but it didn’t work properly and was extremely overbroad) the few times I tried writing a story with them, I ended up taking the prompt away from the AI and continuing it on my own. And now I have the rough rough draft of a very short fantasy action-adventure-romance novel that will never be published. Not unless there’s a market for shitty novellas that indulge in escapist fantasies that are somewhat different from the Fabio-on-the-cover supermarket romance trash variety.

Not that my story isn’t also trash, because it is, but I still like it. Maybe I should rework it into a visual novel script?

2 The parasocial relationship aspect of VTubing is still another deep dive that I’m sure a few people have taken already. I don’t know if I’m qualified to address it myself, but it is an interesting subject. Maybe it’s one I should address — not like I’m qualified at all to be writing about AI, yet here I am completely bullshitting about it.

Actually, I do know more about this other subject, since I’ve spent enough time in VTuber chats on YouTube to know that at least a few people are quite serious when they send love confessions and marriage proposals to their beloveds. Then again, that’s always been a thing idols have had to deal with, so maybe nothing’s really changed.

3 You can say this image lacks intent and meaning, but it sure as hell doesn’t lack value: it sold for almost half a million dollars when it was put up for auction a few years ago, probably for its novelty value since it was touted as the first piece of AI-generated art to ever come to auction. I wouldn’t buy it for more than $20 myself, but since I’m not a member of the idle rich set, my opinion doesn’t matter when it comes to these big-ticket auctions.

4 Of course, there’s also a religious aspect to this question, since many people believe that a God-given or otherwise divinely created soul is the most essential part of what makes us human. That’s a debate I don’t feel qualified to get into — I leave it to the scientists, theologians, and philosophers to argue over all that.

5 To complicate matters further: you could argue that this is exactly what we humans do when we create art, since everything we make takes at least some inspiration from past works of art. But there’s usually more to the creation of art than just copying our influences — we filter those older works through our personal experiences and feelings and create something that’s our own, even if it’s somewhat derivative. The same can’t be said for these AI artists, at least not yet.

A review of 13 Sentinels: Aegis Rim (PS4)

Where to start with this game? It’s hard to say, because there’s a lot to talk about here. 13 Sentinels: Aegis Rim was announced all the way back at the Tokyo Game Show in 2015, but it came out late in 2019 in Japan and late in 2020, only a few months ago, in North America. While it wasn’t released to a lot of fanfare over here, anticipation seems to have been very high among fans of developer Vanillaware, known for their unique art style and great attention to detail with earlier titles like Odin Sphere and Dragon’s Crown.

The big question in these cases is whether the game was worth the long wait. I can’t claim I was one of those fans waiting for five years on the edge of my seat. But after playing through it, I can say that if I had waited that long or even longer, I think 13 Sentinels would have been more than worth it to me. It’s not going to appeal to everyone (just like most of the games I write about here, that’s nothing new) but I liked its mix of gameplay styles and especially its characters and story.

Of course, I’ll be getting into all that in more depth here. Before that, there’s one more thing I have to bring up: this is going to be a no-spoilers review. I’m still putting a disclaimer up even in this case, though, because 13 Sentinels is one of those games that it’s best to go into completely blind if possible. If you trust me enough to just take my word on faith (which I don’t expect at all) then here it is: I greatly enjoyed 13 Sentinels and highly recommend it. But not everyone is going to agree with that assessment, and in any case you probably need more than just me saying “hey it’s great the end”, so I’ll get into why I liked this game so much below without dropping any major plot points or character details, because you should discover those for yourself.

I also promise I won’t make any “get in the robot Shinji” references. They don’t exactly fit here anyway.

On the surface, 13 Sentinels is a game about high school students who have to fight city-destroying mechanical kaiju-style monsters by piloting giant mechs called Sentinels. A lot of the gameplay and plot revolve around these battles and the enormous strain they put on their young pilots, both physically and mentally. As with a lot of other “teenagers in giant robots fight to save the world” stories, though, there’s more going on under that surface.

From the very beginning, 13 Sentinels is split between three different gameplay modes: Remembrance, Destruction, and Analysis. Remembrance is the one you’ll likely be in most of the time. It’s the one that looks like the typical Vanillaware game, only there’s no combat in this mode — it’s sort of in the style of an older adventure game, consisting almost entirely of exploration and character interaction. This is where we really get to know our characters and where almost all of the plot unfolds.

One of the 13 protagonists, Juro Kurabe, at school. The colored bars in the upper right indicate topics or actions that your protagonist has yet to consider or carry out in a scene, some of which are prerequisites to moving their story along.

Each of the 13 protagonists in the game has their own story to play through in the Remembrance mode. In the beginning, the game only gives you a couple of characters to start with, but as you advance their stories, the game unlocks other characters that you can switch between as you see fit. At that point, the initially hidden connections between these characters reveal themselves. These connections are not at all obvious at first in some cases, especially considering the fact that our protagonists are scattered throughout time, with a few from the past of wartime Japan and a few from the far future.

For some reason, everyone ends up meeting in the Japan of 1985, where both the battles against the kaiju and the bulk of the story occur. Figuring out how and why they’ve all converged on this point in time and this place is part of the mystery the game presents. Remembrance involves a lot of tracking down and talking to or otherwise interacting with other characters in the course of this story, but we also get directly into the heads of our player characters. The Thought Cloud is an integral part of this exploration section of 13 Sentinels — it lets the player scroll through the protagonist’s various thoughts, which are updated as they make new discoveries.

Ass-kicking delinquent girl Yuki Takamiya takes a break on the roof in the middle of her various thoughts, one of which is “maybe I should drink this juice box.” Getting your vitamins is important.

But you can’t just make progress through exploration and talking to people: you have to actually fight those big kaiju battles by directing your 13 protagonists in their mechs. This is where the Destruction mode comes in. Destruction is a real-time tower defense game, starkly different from the adventure game style of Remembrance both in its looks and style. Taking place during what the game calls the final battle against the invading kaiju, a horde of giant mechanical monsters, Destruction requires the player to direct a strike team of up to six characters in the defense of a giant terminal that itself acts as a defensive mechanism against the kaiju. Defeating all the kaiju on the map typically leads to victory, though a couple of the game’s 31 maps (not counting the first seven tutorials) require the player to destroy a specific powerful target.

The heat of battle. These screens can get a bit confusing with all the enemies, missiles, and lasers shooting off and flying around, but it’s not hard to get the rhythm of combat down.

A lot of your success in battle comes from preparation. Each of the protagonist’s Sentinels can be upgraded using “Meta-chips” you earn both from advancing the story in Remembrance mode and fighting battles in Destruction mode. Around the middle of the game you’ll be able to unlock some extremely powerful weapons to use against the kaiju, both short- and long-range, allowing you to play defensively by turtling around the terminal and using long-range attacks or offensively by taking the fight directly to the kaiju and punching them in the face. There are four types of Sentinel to choose from as well, each types with its own strengths and weaknesses, so you can mix things up depending upon your preferred play style with a combination of defensive and offensive tactics.

Natsuno’s Missile Rain is stupidly powerful, and I relied on it a lot. Thanks Natsuno, and also Tomi and Keitaro who have the same ability — you saved the team more than a few times. Also god damn, those Sentry Guns.

Finally, there’s the third mode, Analysis, which isn’t so much a gameplay mode as it is a giant cache of information that grows as you progress through the game. Analysis includes a library of previously played scenes that you can return to watch as many times as you like as well as a set of “mystery files” that are unlocked and added to as the game progresses. These files contains information on just about everything in the game, from the characters and their backgrounds, stories, and relationships down to various foods and drinks your characters consume during their adventures. It might seem weird to have entries for such trivial information, but in this game, sometimes the most seemingly trivial bits of information can be important in strange ways down the line.

I’m not kidding; there’s even an entry for strawberry crêpes. Though I have to give credit: the artist made all this food look amazing. I got hungry playing 13 Sentinels a few times for foods I can’t even obtain where I live. Thanks a lot for that, Vanillaware.

It might seem a bit strange at first to put so much information about the game’s major plot points and characters into a library like this. But it doesn’t feel at all like a lazy shortcut to make telling the story easier. On the contrary, I think this Analysis mode is necessary, because the story and its characters’ relationships get so complicated that it’s sometimes helpful to go back and check on a few already established points. Naturally I can’t give any examples without spoiling things (I even went to the trouble of redacting the above image in five seconds in Paint; a lot of work, I know.) It’s enough to say that this mode is very useful in a game like this in which each character’s story has its own flowchart practically.

That takes me back to the story itself. I think the greatest strength of 13 Sentinels by far is in its writing, in the plot and its massive tangled web of characters and relationships. The story is ambitious, but unlike some other works that try this sort of thing and get lost in technobabble and confusion and end up a mess, 13 Sentinels keeps it all together. Part of this might have to do with the organization of the information you receive in Analysis mode and in the character timelines that let you track your progress and jump around from point to point to a limited extent.

For example, should you get crêpes or ice cream after school? This really is one of those branching path decision points.

However, I think more of it has to do with the strength of the game’s characters. Each of the protagonists along with several important side characters are given enough screen time to establish their personalities and motivations. Through their story paths in Remembrance mode, we come to understand how and why they end up in these giant mechs fighting kaiju. These aren’t a bunch of cardboard cutouts either. Each character feels pretty well fleshed out and realistic, allowing the game to build believable rivalries, friendships, and romances. And there are romance subplots in 13 Sentinels, and even though I’m about as unromantic as it’s possible for a human to be, they worked for me — they’re not just shoehorned in for the hell of it but actually play their parts in the larger plot.

That’s love, man, who knows

Then there’s the other half of 13 Sentinels, speaking in terms of gameplay at least: the RTS tower defense section. This one seems a bit controversial. I’ve barely played any tower defense games before, so I really have nothing to compare the tower defense element in 13 Sentinels to, but I found it to be pretty fun for what it was. It was definitely the lesser of the two sections for me, though. Combat in this game, with the exception of maybe two or three boss battles, presented no challenge at all — once you figure out how to play defensively and get the skills to beat the shit out of kaiju without them getting anywhere near you, you’ll be all right for the most part. The second-to-last fight did give me some trouble, but I still beat it on my first try, and I’m not even very good at this sort of thing.

I’m not sure if this game will satisfy hardcore RTS/tower defense fans because I’m not one of them, but playing on hard mode is a good idea if you’re looking for more of a challenge. It’s probably also important to note that, as you can see in the battle screenshots I posted, the combat takes place in top-down view, as if the player is controlling everything from a command center. If you were hoping for the combat sections to be all drawn and animated in that Vanillaware style like they were in Odin Sphere and Dragon’s Crown, you won’t get that in this game. But I didn’t mind too much — the real draw of 13 Sentinels for me was in the story and its interesting character relationships and conflicts.

I wish I lived in Shu’s apartment, what a view. This is my ideal living space.

There’s plenty of style in this game as well. The art is very impressive, much of it handpainted and animated in the typical Vanillaware fashion. That’s one of the reasons I used so many screenshots here, probably more than I normally would in a review like this: the beautiful art adds a lot to an already great experience. The soundtrack is also excellent, from mood-setting pieces in the Remembrance sections to tense battle themes in Destruction mode. And as an added bonus for western players, the NA release features both Japanese and English dubs, so you can choose whichever one you like. Kids these days really have it easy — I remember when we didn’t have that option.

I have more I can say about 13 Sentinels, but not without getting into spoilers, so I’ll leave it there. It’s obvious by this point that I really liked this game and that I’d highly recommend it, especially if you’re a fan of other weird sci-fi material like the Zero Escape series and Steins;Gate. I’ll only add the caveat that it might not be much of a tower defense game if that’s really what you’re looking for. But again, since I’m no expert in the tower defense genre, I can’t say much about that. Again, it’s really all about the story for me in this case, and I was happy with what I got out of 13 Sentinels in that respect. Now I just have to track down some yakisoba pan to see if it’s really as good as the game claims it is, and it will get a perfect 10 out of 10.

Deep reads #3: Just a little broken (Planetarian)

What’s the line between human and machine? If an artificial intelligence were created that seemed so natural and lifelike that we treated it as human, would there effectively be no difference between that artificial life and a natural one? And do these questions even really matter?

If there’s a mandatory reading/playing list of visual novels, Planetarian would have to be on it. First released in Japan on PC in 2004, this kinetic novel has gotten both fan and official translations on several platforms and is now widely considered a classic of the medium, and rightly so. This isn’t my favorite VN, but it is one I enjoy and respect a whole lot, and it takes on the above questions in a unique and interesting way.

Before I begin my look at Planetarian, however, I want to lay out exactly what approach I’m taking with it. I initially tried to write this as a normal review, but then I kept writing until I had a whole damn treatise on the thing. So it’s full of spoilers, both for Planetarian and a certain popular sci-fi film with some surface similarities that I contrast it with, one that took a promising premise and managed to completely shit it up in its last ten minutes (and one that was marketed partly through a harebrained scheme using a fake Tinder profile to catfish SXSW attendees. Okay, it’s Ex Machina.) So if you want to go into either of these raw, here’s your warning.

I do want to persuade people who haven’t experienced Planetarian yet to check it out, though, so here’s a one-sentence no-spoiler review: if you like the idea of a short post-apocalypse sci-fi story with excellent characterization, voice-acting, and music, but no branching decision points or route because it’s a kinetic VN, you should like it. I think the ending of this work is pretty well known by now since Planetarian has been around in various forms for 16 years, but I still feel the need to put a warning up here. It’s only a few hours long anyway, so it’s not a huge time investment.

The basic premise of Planetarian is that the world has gone completely to hell. About thirty years after a nuclear war and its aftermath destroyed almost all of humanity, Earth is only inhabited by still-operating autonomous weapons and a scattering of human survivors doing their best to live off of the ruins of their dead civilization. A constant radioactive downpour simply called “the Rain” makes this new world even more difficult to live in. In the midst of all this misery is our unnamed protagonist, simply called the Junker, a man who makes a living off of salvaging useful scraps from the old world to trade with: parts, food supplies, and the extremely rare and valuable preserved packs of cigarettes and bottles of liquor. Junker is exactly the kind of protagonist you’d expect to find in a post-apocalyptic work like this. He’s tough and battle-hardened, always armed and on the lookout for valuables and potential enemies, both mechanical and human.

At the opening of Planetarian, Junker has come across a “sarcophagus city”, a settlement that has been heavily fortified against attack. Unfortunately, those defenses weren’t quite enough: the city was abandoned by its population long ago, left to become yet another ruin. This is an opportunity for Junker, who thinks he may be able to salvage some useful items here.

There is one other being still operating in this dead city. Her name is Hoshino Yumemi, a robot built in the form of a young woman. Despite the fact that the city had been long since destroyed and emptied of its population, Yumemi still works for exactly one week per year as the receptionist, usher, and hostess of the Flowercrest Department Store’s planetarium, spending the rest of her time in sleep mode charging at a station that’s still working off of a trickle of power somehow still available from a nearby vacant military installation. Since the outbreak of the global war and the exodus from the city, however, the planetarium hasn’t seen any business — not until Junker arrives there looking for shelter.

Junker is shocked to find a young woman alone in this ruin and immediately suspects a trap, but he soon realizes that Yumemi is just a robot who has been operating autonomously all this time. As Yumemi herself explains, she was left in charge of the planetarium while the human staff were out. Since the day they left the city almost thirty years ago, she has carried out her duties to the best of her ability for the one week per year that she’s able to operate. And what luck — she happens to be freshly recharged and active when Junker arrives. Yumemi, seemingly oblivious to Junker’s appearance and all the destruction around her, processes him as a customer, greets him warmly, and tells him that in honor of his status as the 2,500,000th customer the staff has prepared a special projection that she intends to show him. She then offers him a makeshift bouquet made of wires and junk she found lying around, apologizing profusely and explaining that the florist’s shop downstairs had unexpectedly closed for the time being. She also admits that he’s not really the 2,500,000th customer, but she’s rounding up because there hasn’t been much business lately.

Junker naturally does not give a shit about any of this. After trying without success to explain to Yumemi that he’s not a god damn customer, he lays out his supplies and equipment to dry, then drifts off to sleep in one of the planetarium’s seats. When he wakes up, Yumemi is still around performing her duties, and she cheerfully greets him, addressing him as “Mr. Customer” (or okyakusama, a term like “honored guest” that doesn’t quite translate because we don’t have a similar term in common use in English) and doing her best to serve his needs. Of course, Yumemi can’t really serve Junker’s needs. When she offers him a refreshment, he asks whether she has any liquor in sealed bottles, and she tells him there are liquor shops on a lower floor. Tragically, that lower floor is completely flooded and inaccessible, so Junker can’t even have a nice drink to calm him down.

Yumemi continues to insist that she’ll show Junker the projection, and he finally gives in to her demands if only to shut her up. However, there’s a problem: the projector is broken. No big surprise, since the planetarium has been inactive for nearly 30 years, but Yumemi is seriously distressed when the projector doesn’t move or respond at the start of the show. Since she was built to be a sort of greeter/hostess and not a maintenance worker, there’s not much Yumemi can do to fix the giant machine, and so she asks Junker if he can repair “Miss Jena” as Yumemi refers to it.

This leads to the first of two fateful decisions Junker makes. By deciding to help Yumemi out, Junker takes up valuable time and energy that he admits he should be using to get the hell out of the city and resupply. He’s established that the planetarium and attached mall don’t have anything of value to him. Yet he stays and starts working on Jena, an extremely complex piece of equipment with a bunch of small moving parts that hasn’t been maintained for three decades. Meanwhile, Yumemi can only stand by and express her concern. She clearly feels bad about asking a valued customer to repair one of the planetarium’s machines and tries to help Junker by asking him if various tools might be useful, but it’s obvious she wasn’t designed for that sort of thing, so she steps back and lets him work.

After a couple of days of work, Jena is finally repaired, and Yumemi is able to run the special projection she had planned. Junker is still anxious to get the hell out of there, but once the lights dim and Yumemi starts her presentation, he’s drawn in. So much so that when the power fails for good shortly after the projection starts, Junker asks Yumemi to continue her monologue as he closes his eyes and uses his imagination to fill in the visual gaps.

If you’ve read Planetarian already, this may seem like a weird statement, but this scene provided the biggest emotional punch for me as Yumemi talks about the birth and growth of the human race and of its reaching out to the stars through the space program. The same space program that was in progress when the global war began 30 years ago, destroying its base on the Moon, grounding its spaceships, and and eventually killing the great majority of humanity. It’s all the more heartbreaking because, despite the fact that she’s a robot, Yumemi seems genuinely proud of humanity’s growth, just as though she were human herself. But her information is painfully outdated. Junker knows the truth of the matter all too well, but he lets Yumemi finish without saying anything about it.

When the show is over, Junker is ready to leave. But not without Yumemi. This second serious decision puts Junker at yet another disadvantage — Yumemi doesn’t seem to understand how dire the situation is outside the mall and planetarium, and she’s already told Junker that she’s not designed to handle rough environments or to move very quickly. Junker nevertheless doesn’t want to leave her there, and presses the facts on her that the planetarium won’t recover its limited source of power again and that she’ll never see another customer show up. Yumemi still seems optimistic despite Junker’s warnings, so when she offers to walk him to his car, a sort of post-apocalypse combat vehicle, he takes her up on her offer and decides to bring her along with him to a nearby inhabited settlement.

Getting to his vehicle is no easy matter, however, and it’s even more difficult when he’s essentially doing an escort mission. Yumemi trips several times and admits that she hasn’t been very well maintained lately. But she still keeps her spirits up, pointing out popular restaurants and attractions around town and printing coupons from the port in her ear for him to use, apparently not recognizing that that they’ve all been long abandoned and lay in ruins. Eventually, after several breaks to let Yumemi recover and prevent her from overheating, they reach the city wall, close to Junker’s car. A giant tank with a massive gun sits at the entrance of an opening in the wall through which they’ll have to pass, but Junker believes trying it would be suicide — despite the end of the war, the automated weapons deployed back then are still active and will attack anything that moves. So Junker tells Yumemi to hang back in a relatively safe place while he tries to destroy the tank with a grenade launcher.

Junker’s grenade is unfortunately a dud, and the tank turns its gun on him. He manages to escape and mostly disable the machine against all odds in the game’s only action scene, but it’s still barely functional and is about to kill Junker when Yumemi steps between them in a dramatic Tienanmen Square moment.

Yumemi tries to send the tank an electronic signal to get it to stop attacking, but in its final moments it shoots its gun directly at her.

Yumemi is torn apart at the waist, but she’s still able to function for a few minutes, just long enough to show Junker some of her memories recorded in her eyes: of happy guests, adults and children, telling her how much they enjoyed their time at the planetarium, and of the rest of the staff being forced to evacuate the city and saying their painful goodbyes to her. She then reveals that she realized long ago the planetarium was finished, but that she was happy to see one more customer show up. As she finally shuts down, Yumemi opens the port containing her memory card, and Junker takes it and seals it in a waterproof case, resolving to find a new body for her somehow so she can live again.

And that’s Planetarian. Quite a sad story in typical Key style — this studio is well known for creating melancholic visual novels. As miserable as the whole thing might seem, though, the story of Planetarian is not a hopeless one. Yumemi’s body is destroyed in the end, but her mind essentially lives on, waiting for Junker to find a new vessel for it.

What’s more interesting to me than the ending is the relationship created between Junker and Yumemi, a human and a robot. From the beginning it’s no secret that Yumemi is not a human, and a lot of her mannerisms reinforce that. When asked a question she doesn’t know the answer to, for example, she’ll tilt her head a bit and then deliver word-for-word the same response about not being able to make contact with some control center that she’s programmed to message in such cases. Her insistence upon carrying out her regular duties in a workplace that’s clearly been abandoned and left to rot for thirty years also seems kind of inhuman. A human would have left the planetarium behind long ago, just as Yumemi’s coworkers did, but she keeps performing her programmed duties faithfully.

But there are things about Yumemi that also seem strangely human. One of these is her extreme talkativeness. Yumemi simply won’t shut up. Junker is clearly annoyed by this and tries giving her a command to stop talking — a command that she acknowledges for about ten seconds before breaking it and asking him a question, after which he gives up trying.

Yumemi explains that this chattiness is caused by an error in her programming, one that was never fixed because the staff of the planetarium thought she was cuter for it. She refers to herself as “just a little broken” both because of that design flaw and her recent lack of maintenance. Certainly, Yumemi doesn’t act like a perfectly honed android of the kind you might see in some other sci-fi works, but these imperfections made her seem all the more human to me. She also constantly shows genuine concern for Junker despite having just met him, asking if he’s feeling sick and offering to call the mall’s medical center that she doesn’t realize is now abandoned. Indeed, Yumemi seems determined to help Junker out and tend to his needs as the “customer” he is, even when he insists he’s not one.

Considering all this, it’s not a great leap for Junker to start thinking of Yumemi as less of a machine and more of a human, at least in terms of how he treats her. The pair have the kind of chemistry where one complements the other — Junker’s bitter, harsh, practical attitude with Yumemi’s optimistic and cheerful one — and they start to have real conversations by the end of his stay at the planetarium. The first time I read through Planetarian, I thought it was a bit weird that this extremely pragmatic guy would decide to bring a slow, partially broken robot along with him through the streets of the city, where autonomous, heavily armed tanks were still operating. Junker wonders that himself and doesn’t seem to understand exactly why he’s doing it. But there has been a connection created between the two when the final part of the VN begins, to the point that I can believe Junker simply couldn’t allow himself to leave Yumemi alone in the now unpowered mall to shut down — effectively to die, left to be “harvested” for her parts by other junkers as he puts it.

This is where Planetarian totally departs from a lot of other modern sci-fi. When I watched the 2014 film Ex Machina a while back, I couldn’t help but be reminded of Planetarian and how completely different each work was in spirit, despite the fact that they both deal with human/AI interaction. Ex Machina takes place in a near-future Earth that’s still thriving, in which the eccentric genius CEO of a massive search engine company has built a line of realistic androids. Said CEO rigs up a fake contest to select one of his employees, a coder/programmer type named Caleb, to spend a week at his high-tech, high-security mansion in the wilderness. There Caleb gets the chance to run a series of tests by having conversations with Ava, like Yumemi an android in the shape of a young woman. Ava seems to be curious both about the outside world, which she hasn’t seen, and about Caleb himself. She also comes off as having an almost human-seeming sense of humor and a pretty sharp wit. After a few days of testing, Ava tells Caleb that she knows he’s attracted to her, that she’s attracted to him, and that she wants him to help her break out of the CEO’s mansion and escape.

Despite their efforts to conceal these parts of their conversations, the CEO realizes what’s going on, but in a double-twist Caleb reveals that he outsmarted the CEO by secretly fucking with the power system so that he’d be sealed inside his own high-security bunker of a house without being able to get out while Ava and Caleb would run away together. CEO tries killing the plan by ordering Ava to go back to her room, but she and another android get the better of him in a fight and stab him to death. We’ve seen him act like a real asshole to them throughout the film, so sure, this makes sense. However, in a final betrayal, Ava traps Caleb in the house and escapes without him, leaving him to die as well. The end.

Does this remind anyone else of those old creepy Svedka ads? Is it just me?

What message is to be taken from Ex Machina exactly? Caleb admittedly didn’t think through his actions fully, but he was motivated by a desire to help Ava escape because he essentially saw her as human, or at least as a being deserving of human rights. While Caleb did mean to seal the CEO into a virtual tomb in the course of his plan, he also found and watched tapes of said CEO treating the androids like garbage during tests and generally being a dick, and he also knows of his plan to erase Ava’s memory at the end of this testing phase. So his feelings are a bit understandable. However, the relationship Caleb thinks he has with Ava is pure fantasy. She’s been manipulating him this whole time, and far from being grateful for his help, she traps him and effectively murders him at the end of the film for no clear reason that I can understand, other than director/screenwriter Alex Garland wanting to throw a final twist in to shock us.

At first, Ex Machina left me asking “so fucking what?” The actors are good (Domhnall Gleeson and Oscar Isaac as Caleb and the CEO, for you Star Wars sequel trilogy fans if there are any left, and Alicia Vikander as Ava) and the look and feel of the movie in general are pretty nice, so I can’t exactly call it total garbage. But the writing. The first 100 minutes of the movie now seem entirely pointless, with its attempts at making me feel bad for the plight of Ava by making her come off as self-aware and sympathetic — such a being would have at least recognized Caleb as her ally and let him live, even if she’d been manipulating him up to that point. But no, turns out she’s nothing of the sort, more of a HAL from 2001 sort of character. Only 2001 took the time to establish HAL as a scary psychopath sort of AI making the course of the story believable, whereas Ex Machina just throws us an ending twist without bothering to set it up in the slightest.

So the message I’m forced to take from Ex Machina — because there clearly is a message in there; everything about the film suggests it’s meant to be taken as Serious Art instead of a basic horror movie — is that we can’t trust those god damn androids because there’s no way they’ll treat us with any care or affection despite what we might think. This is a depressingly pessimistic message. That’s fine with me; I’m a depressingly pessimistic guy myself, so I get that.

But what I can’t forgive is the sheer dishonesty of it. Ex Machina presents a dark future without any real argument to back it up. While many critics and fans have praised Ex Machina, I believe Garland completely screws up its treatment of its central human/AI relationship, which is quite an unbelievable and stilted-feeling one created to express the message, when the message should instead flow naturally from a believable story. Planetarian also depicts a dark future for humanity in its global war and post-apocalyptic setting, but in creating the relationship between Junker and Yumemi it doesn’t try to pull a cynical trick on the reader. Yumemi is exactly what she seems from the first time Junker meets her — an android who likes the company of both her machine and human colleagues. She has fond memories of working with the planetarium staff and helping customers and has a desire to continue her work.

Whether that’s because she’s programmed to do so doesn’t seem to matter anymore, at least not to Junker. By the end, he doesn’t see her as a mere piece of machinery. Yumemi herself, while conscious of the fact that she’s a robot, doesn’t want to be separated from humans. This is the meaning of her saying “please do not divide Heaven in two”, one of the game’s best-known lines — when she asks whether Junker has ever prayed to God, a weird sort of theological question comes up about whether there might be a separate God of Robots. Yumemi says her coworkers told her that robots get to go to their own Heaven when they shut down, but because she doesn’t want machines to be separated from humans, she prays that they can all go to the same Heaven.

This is where I think you can find the optimism in Planetarian. It’s a sad story with a bitter ending, sure. But there is hope in the end, both for Junker and Yumemi, and maybe for both humans and machines beyond them, living in the world together. Yumemi sees both organic humans and other, non-humanoid machines like Jena as her friends and colleagues, and she even says Junker shouldn’t blame the tank for what it did — in the end, it was simply doing its best to carry out its duties faithfully.

This view is very different from the one given by works like Ex Machina, in which humans create technology that ends up destroying them of its own will. In those works, there’s an assumption that any form of advanced AI will necessarily be separate from the natural world. Humans are animals, androids are machines, and there can’t be any meaningful emotional relationship between them. When a well-meaning character like Caleb foolishly believes he’s created one with an android like Ava, she ends up betraying him. She can’t empathize with him, and he was stupid to think he could empathize with her. Planetarian, by contrast, does not make any such assumptions. Humans started the global war that wrecked civilization. They used technology to do it, but the story doesn’t give any indication that the AI employed in the war rebelled against their human creators or did anything other than follow the orders given to them.

I’m not saying a robot apocalypse will never happen. But it seems both disingenuous and lazy to just assume that advanced AI will definitely turn against its creators when you’re putting together a work of fiction, or that they’ll even necessarily see themselves as that different from their creators.

I wrote at the start of this post that Planetarian isn’t my favorite visual novel. While I don’t have any problem with kinetic novels, I prefer VNs that give the player dialogue and action options and branching story paths. And I don’t know if writer Yuichi Suzumoto is responsible for this or if it’s the translation, but the prose occasionally gets really awkward — just see the above screenshot for an example. Thankfully it doesn’t happen that often, but those instances stick out and hurt an otherwise good game.

But I’d still rank this pretty highly among the VNs I’ve played. A good story can end with disaster and total despair, but the way it gets to that ending is important. Planetarian doesn’t take the same straightforward, lazy “technology is bad” route that Ex Machina and many other modern sci-fi works go with. And it’s not afraid to express the hope at the end that maybe things won’t be so miserable one day, and what the hell is wrong with that? Nothing. In real life, people keep hope alive even in the worst of circumstances, so it’s not a sin to give your audience some hope as well, despite what some writers and directors seem to think.

And that’s true even if that hope directly follows a tearjerker scene. I mean, I didn’t cry when Yumemi got blown up. Really, I didn’t. I just had bad allergies that day. You know how that pollen is in the spring.

***

I hope I’ve represented Planetarian well enough here. It also has anime OVA and film adaptations that I haven’t seen, but I’ve heard good things.

I also want to note that I’m not trying to do a “western vs. Japanese take” comparison with this commentary. Reading back through it, all the crap I dumped on Ex Machina might make it seem that way to some people, and everyone knows I’m a degenerate weeb after all, but it’s not the case. I only meant to highlight two approaches sci-fi writers have taken with regard to human/AI relationships and how I think one is more natural and honest than the other. If you want proof of my sincerity, here you go: the Spike Jonze movie Her does thousands of times better at this than Ex Machina, and it involves an actually believable romance between a human and an AI character if that’s what you’re looking for. 𒀭