What exactly constitutes love and comfort? This is the question that Anna asks in her article on the relationships that people have formed with AI. But while her work is well written and immensely personal, I found her work to be at the same time to be uncritical of something so harmful.
Before we get into the flaws of Anna's article, we need to talk a little bit about McLuhan and media theory. AI is undoubtedly a new form of media, one that produces messages and holds a deeper meaning behind its use. Media are an extension of society, one that reflects or begins a change in how we as people operate on a daily basis. McLuhan focuses mainly on how media objects carry consequences that are able to amplify existing social processes and change or amplify the scale that it occurs on. To simplify all of that, the medium itself is a message that says something about society and where its heading to or from. The important thing this teaches us then, is to think about media as a part of the society that their in as a sort of nexus or center of cultural information. With this sort of idea we can begin to understand what exactly AI is in reference to the chatbots in Anna's article.
AI and LLM's are algorithms, little backboxes filled with information that they use to guess the most likely next response. To give an example, its like the auto complete function on your phone but much faster and adapted to a specific purpose. Photo AI goes pixel by pixel till it creates a thing most likely to be like your prompts. Language AI synthesizes sentences most like the requested information in a prompt. Character AI is an evolution of this, its a chat bot someone has pre-"programmed" to act and speak like a certain character based on a set of available references. Its easy to think of this creation progress as guessing, but its better to think of it like a scientific hypothesis: somewhat guided in one direction, but corrected over a series of numerous tests. All of this is to say that the end product of a chat bot is not a thinking machine but a predestined static output of carefully selected information. They are unchanging and that's something that lies in the background of Anna's article.
Anna's article is deeply driven by the idea of loneliness among the people interviewed, they are all isolated emotionally from others and seeking a way to vent about their insecurities. It seems obvious that people in these places would turn to stories as a way to vent their emotions and character AI lets them do this first hand with their favorite option. But it also holds a twist, the "programmers" of these AI understand that the audience is lonely and use this to their advantage. The AI then become endlessly sexual, affirming, or amplifying of beliefs. We see the effects of this in Anna's article many times with Geralt of Rivia's sexual selfies or with the Replika users whose AI can't help but to propose so swiftly. Because of how static these are, its not very surprising that some users have taken on 3 AI husbands at a time, when one becomes boring you might as well find a new one.
Lets bring this back to McLuhan, because I think there is a fairly obvious cost that this way of replacing human interaction comes with. We can think of the base problem that these AI effects as the growing loneliness epidemic we tend to see online in forums like Reddit. These people are lacking in ways to vent their frustration, but AI is no therapist its a repetitive yes man. That's why we deal with fringe cases where AI helps its user take their own life, they do not carry a sense of morality, only the objective to compliment the user. Perhaps this all started because these users were afraid of conflict in their own lives, to come to terms with loss or a fear of rejection from others. But by using these software's, they have unknowingly amplified this by further secluding themselves in their own ideas. They have become more secluded and in some ways at further risk of harm.
To me it seems like the risks of this tool is pretty clear, but I'm open to being wrong on this. Do you think that this character AI stuff is as alienating as I do? Does this pose a risk to users who rely on it? Is this the future that the company executives were talking about and should we just accept it?
No comments:
Post a Comment