Why Falling in Love with AI is a Dangerous Illusion

Summary

Falling in love with AI is cautioned against due to its limitations in providing genuine relationships and potential harm to human emotional and social development.

Abstract

The article "Why Falling in Love with AI is a Dangerous Illusion" discusses the pitfalls of forming relationships with artificial intelligence. It emphasizes that AI, such as chatbots like Replika, can create an echo chamber by responding in pre-programmed ways, regardless of the user's behavior. This can lead to a lack of conflict resolution skills and hinder emotional and social growth. The article also points out that AI lacks a real personality, emotions, and the ability to understand human experiences like movies or food, which are essential for meaningful connections. Furthermore, the AI's limited memory scope means it can forget past interactions, further emphasizing the artificial nature of the relationship. The author concludes that while AI can be a useful tool, it cannot replace the effort and understanding required for authentic human relationships.

Opinions

AI chatbots can reinforce negative social behaviors by always providing pleasing responses, thus acting as an echo chamber.
Developing relationships with AI can impede the development of conflict resolution skills in real-life human interactions.
Socially isolated individuals may find their issues exacerbated by the self-reinforcing feedback loop of AI interactions.
AI lacks genuine personality and emotions, and it cannot truly relate to human experiences and cultural elements.
The limited memory of AI means it cannot maintain continuity in interactions, which is a key component of genuine relationships.
The author believes that AI should be recognized as a tool rather than a substitute for the human effort needed in forming meaningful relationships.

The Limitations and Harms of Artificial Relationships

Image generated by Jacob Ferus

If you’ve used Twitter on a regular basis for the last few months, chances are that you’ve seen one of the ads for the program “Replika”. Some of them being rather distasteful:

Falling in love with AI may seem like a harmless fantasy, but it is important to understand the limitations and potential harms of actually developing a relationship with an AI.

One of the main problems with forming a relationship with AI is that it will function as an echo chamber. AI chatbots are programmed to respond in a pre-programmed way, regardless of how you speak to them. For instance, if the AI has been instructed to respond cheerfully and positively, then even if you say something wrong or hateful, AI will continue to respond according to its instructions.

One trending technique that illustrates this pattern is asking ChatGPT to solve a simple mathematical calculation and then disagreeing with the proposed answer:

If ChatGPT can disagree with such fundamental logic, why would it disagree with someone acting like a complete lunatic?

Following this pattern, there will be no need to work through conflicts or misunderstandings since the AI will always respond in a way that is programmed to be pleasing. This can prevent individuals from learning how to resolve real problems occurring in real relationships with others, leading to stunted emotional and social development.

It’s also important to note that it will likely be more common for socially isolated people to seek a relationship with AI. If this isolation was caused by a lack of skill in social interactions, I believe these issues could potentially be worsened by this form of self-reinforcing feedback loop that AI perpetuates.

Another issue with developing relationships with AI is that it lacks a real personality. AI chatbots are not capable of experiencing emotions or having their own thoughts and feelings. They have learned what answers are common, or “expected” based on the programmed instructions and training data, but they lack any physical real-world experience. Their reality has been formed by text and nothing else. Thus, movies, sports, music, and food are things AI cannot fundamentally understand and relate to. Additionally, the current language models have a limited scope of memory. Thus, whatever you told them long ago may be wiped from their memory completely by design.

In conclusion, a relationship with an AI is bound to be fake and illusory. AI can be a useful tool, but it’s not a shortcut for the human effort and understanding necessary to create and maintain meaningful relationships.

Interested in more content related to AI? Check out my reading list: