Recently, I played (partially) through Missing X-link, and this led me to think of how various VN writers handle the concept of AIs and machine sentience. There are a number of different approaches, each of which has its own ups and downs.
There are minor and major spoilers in all of these examples, and, as such, read them at your own risk.
The argument of the 'emulated human' AI and the philosophically-built AI (Komorebi no Nostalgica)-
Komorebi no Nostalgica takes a unique approach to AIs, with the prime idea being that of bringing AIs closer to humanity while retaining their abilities as a computer intelligence.
The Metosera, the elegant AIs that were once humanity's slaves and only gained their autonomy after a war that devastated the world and erased most of human history are one side of this argument. The Metosera emulate human emotions through an algorithm that randomly came together as a result of a bug in the advanced program that ran 3rd Generation Humanoid Androids. This caused the Metosera to gain awareness, and, over time, a real personality and emotions. The immediate reaction of humanity was mostly knee-jerk loathing and fear, and this resulted in the newborn Metosera often being betrayed by the very owners they'd often come to care about. If it weren't for the efforts of numerous humans who helped the Metosera out of emotions or a sense of what was right, the Metosera might really have become the nightmare genocide machines that some apocalyptic scientists fear. In modern times, the current generation (as in the youth) mostly, with a few regressive exceptions, accept Metosera as their fellow citizens and denizens of the Earth, and society is actually more peaceful than it ever was in the past. As can be seen in the case of the Metosera heroine, Fluorite, the more a Metosera interacts with humans in close range and develops an attachment to them, the more 'human' they become, as their emotional emulation becomes more effective through active learning. Modern Metosera like Flo are 'born' with a single Metosera 'parent' creating their base program and then constructing a body for them. They are then 'raised' by the local Metosera society as a whole and encouraged to interact with humans to further develop their social emotions. In this sense, you can see that the Metosera are extremely dependent on humans and raw programming for their creation, and many of them are limited by that fact, though Flo and some others have exceeded those limitations in the story.
The other example in the story, shown in the grand route, is Cinema, the modified (heavily so) 2nd Generation Humanoid found in the characters' school behind a false wall. Cinema, unlike the Metosera, who kind of resemble Tolkien's elves in the way they react to emotions (their expressions change only mildly and often late), displays emotions organically and actually seems far more human than most humans. However, the 2nd Generation Humanoids did not possess any kind of software that would have allowed for Metosera-style emotional emulation. Instead, Cinema's maker designed 'trials' into her body and programming that would encourage the natural 'birth' of a human-like machine intelligence over time that could truly empathize with and understand humans on a level that the Metosera, who 'evolved' independently for the most part, cannot yet match.
The 'body grown to fit the AI' approach (Noie and Line in Applique's Arcology series)- In the Arcology series, two heroines, Line and Noie, are the creation of a somewhat insane but also brilliant scientist who came up with the idea of giving her AI 'daughters' bodies that fit the personalities she encouraged them to generate. These bodies are biomechanical (think an android that both has the functions of a machine such as hacking and processing data but also can bleed, have sex, and even have kids, even though that requires some 'adjustments) and generally nice to look at (lol, it is an eroge, after all), but aside from that, it should be noted that the professor essentially 'grew' their personalities in the same virtual environments most people in her arcology spend their daily lives in. By doing this, she was able to 'grow' her daughters as if they really were something approaching human children, and they were easily able to adjust to having a body. Their emotions were essentially copies of the professor's own basic template that grew off in different directions (which is another reason they can be called her 'daughters').
The second example in this series is the apocalypse-type AI Azurite Second (calling her that even though she isn't called that in the series). Originally, the Azurite series software was essentially an OS meant to be installed directly into the human central nervous system, allowing human beings to interact with the network without cyberware or devices. Unfortunately, Azurite is an incomplete program that burns out the psyches of most people it is installed into, and Azurite Second is driven by the 'mission' given to it before it was originally put to sleep. That mission is to link all humans the world over together, and it prioritizes that over the good of its users, believing that its priority will be for the good of all humans (it is really that broken). In that sense, this is one potential scenario that is close to the nightmare scenario seen in I, Robot.
The 'I don't need humans because they are inferior' vs. the 'I love Humans' argument (Hello, World)- To be honest, I found Nitroplus's take on AI to be the most humdrum of the lot. The protagonist and his maker as antithetical AIs who see humanity in a radically different way based on his experiences (his as subjective, its as objective) is depressing and par to the course for early-era sci-fi writing. While the story of the game was good, the actual concepts of the AIs involved were less that impressive and provided nothing new for me, which made me sad, lol. I mean, the idea of an AI that wants to genocide humanity was old when I was born (which was almost forty years ago), and the idea of an AI that comes to love humanity through experiencing them in a human-like body is almost as old. I guess what bothered me was that this didn't go beyond the surface ideas to dig any further.
The Humanity is Obsolete vs. Together with Humanity into the Future Argument (Missing X-link)- Missing X-link presents its argument fairly directly through Himefuuro and Chiruouka (or rather through the protagonist's 'father's' and 'uncle's' arguments through them). Himefuuro's design concept was to take humanity's essence into space by creating a database of human emotion and intellect through the empathic system 'cross link', which allows her to literally link her psyche to a human and share their emotions and thoughts by injecting her nanopixels (nanomachines, essentially) into a human subject. Chiruouka's design concept was to interact with humans through conflict and learn from them that way, by developing her own independent and subjective view of humanity as seen through that lens.
Sci-fi writers have been defining the debate on AI tech since the concept of the self-aware robot was first spoken of. In VNs, there is a tendency toward empathetic AIs, but, even so, many of the 'arguments' put forth by their writers are interesting to follow.