Comments
55
Quick Intro: My name is Strad and I am a new grad working in tech wanting to learn and write more about AI safety and how tech will effect our future. I'm trying to challenge myself to write a short article a day to get back into writing. Would love any feedback on the article and any advice on writing in this field!
People are getting nervous about job loss due to AI. AI is progressing fast, massive layoffs are sweeping the nation, and many leaders in AI are racing towards artificial general intelligence, AI that can do everything humans can do but better.
While the timeframe for when AI will be able to take over most jobs is highly debated, the idea that, in principle, AI could reach the point where it could do most, if not all, jobs better than humans seems widely accepted. This has gotten me thinking about whether their will be any jobs left where people would prefer humans over AI.
In other words, if the only difference between a job done by a human vs an AI is ones knowledge of whether the job is being done by a human or AI, would there be any jobs where people still prefer humans?
Press enter or click to view image in full size
In order to answer this question, I turn to human relationships as an aspect of life where the inherent value of being human seems more potent. Based on my intuition, I have an assumption that a majority of people in today’s society would prefer to have a real, human, relationship (friends, romantic partners, etc.) over one with AI. While its debatable how many people feel this way, I, at the very least, have this preference. Because of this, I think it is worth looking into why someone might prefer a human to an AI for personal relationships and see if any of these reasons could apply to human preferred jobs.
I propose three properties that allow humans to have an edge over AI when it comes to personal relationships. These properties are, a greater sense of consciousness, proof of human experiences, and identity.
Greater Sense of Consciousness
When we talk to another human, we assume they are conscious. We are human and we experience consciousness, and most other humans act as if they have consciousness, so we assume that other humans are conscious. AI, on the other hand, is a different story.
We now have LLMs that can have full, human-like, conversations with us. LLMs can discuss deep, serious topics and even portray emotions and a sense of self. Despite this, many people are skeptical that these LLMs are conscious at all.
It is possible that as AI becomes more intelligent and indistinguishable from humans, more people will start thinking they are conscious. However, I have a feeling that humans could always have a bias towards believing that other humans are more likely to be conscious than AI since consciousness has been ingrained into us as an inherent property of humans.
I believe this bias is a factor in why many humans prefer other humans for relationships over AI. People want to feel like there is someone else with subjective experience on the other side of a personal relationship. That’s why talking to chatbots in today’s age can sometimes feel a bit off. We have a greater uncertainty as to whether any subjective experience is actually happening on the other side of the conversation.
Proof of Human Experiences
As humans, we have an understanding of the typical experiences every human goes through. Being born, growing up through childhood and adolescence into adulthood, having emotions, having to eat, use the bathroom, sleep, etc. We also have an understanding of the constraints we have by being human. We aren’t super intelligent, we can’t fly, can’t breath underwater, etc. As a result we have an understanding of the typical struggles that come with being human.
We also know that other humans, also experience these things. AI on the other hand does not. AI is not birthed in the same way humans are. It doesn't go through stages of life like humans do. It doesn’t experience emotions or the struggle of maintaining a biological body to survive.
An understanding that all humans have a shared set of similar experiences is another factor in why people prefer humans for relationships over AI. Humans love to relate with one another. While an AI can have a great understanding of human experiences and even simulate someone who has been through them, its not quite the same as actually experiencing them.
For example, it can be quite motivating for a human to see another human succeed or get through a hardship. As humans, we all experience similar things, so if they did it maybe we can too. This feeling is harder to replicate with AI since they are made to be super human. It’s hard to see yourself in a being that far exceeds your own capabilities.
Identity
This last property is not specific to humans, but still acts a factor as to why people might prefer some humans to AI for personal relationships. Each human has a unique identity. They have a mix of personality, genetics and history that makes them who they are. When we form a relationship with someone, we are forming a relationship with this identity. There is an inherent value to each identity that can’t be replicated by anyone else. Once you form a relationship with someone, no one else can truly be that person. Sure anyone can tell you about their day, but you likely have specific people in mind that you want to hear from.
Now of course, AI’s can also have identities and as AI becomes more intelligent, these identities will likely become as fleshed out as humans. Despite this, I think it is still very plausible that people will still prefer the identities of some humans to the identities of some AIs. For example, despite being much less intelligent than us, many people would prefer the identities of their pets over random strangers. This gives credibility to the idea that human identities can win out over AI identities despite the AI’s advantage in intelligence.
How the Three Properties Map Onto Jobs
If the three properties I laid out cause people to prefer humans over AI for personal relationships, is it possible that any jobs which rely on these properties might also have a human bias?
I believe, based on these properties, there are two main categories of jobs in which humans could have a competitive edge after the rise of AGI; Jobs that rely on human connection, and art.
Jobs Relying on Human Connection
Jobs that rely on human connection include professions such as therapists and caregivers. While there are technical aspects to these jobs in which AI, in principle, could master, human connection also plays a vital role. I believe this connection relies on a human’s greater sense of consciousness and proof of human experiences.
When working with a therapists, people expose the most vulnerable aspects of their life. Part of what makes therapy work is the perception of another person on the other side listening and caring about ones issues. A perception that could be difficult to achieve if one has doubts about the consciousness of their AI therapist.
Another aspect of effective therapy is the ability therapists have to disclose their own experiences in order to help relate to the patient. When a therapist discloses their own experiences, the patient gets to see another human who has been through similar situations and prevailed. This can potentially foster more confidence in the patient on their ability to get through their own situation. This aspect of therapy is lost with AI since they do not have human experiences to disclose.
The same arguments for therapists can also be applied to other professions such as caregivers and nurses where positive human connection is essential. In situations where human connection is required, the properties of consciousness and lived human experiences are important for the connection to feel real and, therefore, be effective.
Art and Entertainment
There is no doubt that AI will become more involved in art as its capabilities progress. It can already generate images and audio that resemble beautiful paintings and euphoric melodies. However, in addition to the physical look and sound of art, the story and meaning behind art is also very important to its success.
Art is often used to communicate ones own human experiences in a way that others can enjoy. Just knowing that art represents an actual experience that another human had can really allow someone to appreciate it at a deeper level.
While AI generated art might look and sound good, it lacks this communication of true human experience. Even if AI can generate art that represents its own “experiences,” the uncertainty of whether the AI is truly conscious enough to actually experience these things might make it difficult for humans to truly resonate with.
Lastly, even if AI is on par with human art, it is still possible that people will have favorite artists, both AI and human. The inherent value of the identity of these human artists might be enough to keep them relevant and preferred over some of the AI artists.
For example, lets say you are a huge fan of John Green’s work. John Green is an American author and educator who has written many young adult novels that helped him gain a large audience. In the future, there could very well be an AI trained on John Green and his writings which would allow it to generate books perfectly inline with the style of John Green. While I am sure many John Green fans would read these books, and even enjoy them, there would still be an inherent value in John Green himself writing and releasing his own book. An inherent value strong enough that many fans would likely prefer to read this book over the generated ones.
People don't want to just have an accurate idea of what someone they value thinks. They want to actually know what that person thinks. Since art is so tied to the person who makes it, I believe that it will still be possible for people to create good art using their human experiences, and then find and maintain an audience by utilizing the value of their unique identity.
Takeaway
When going through this analysis, I was shocked to find that art and human facing jobs might be the ones to last in a world with AGI. With how much generated art LLMs are already able to create, along with their ability to create some level of connection with people, I would have thought that these types of jobs where the first to go.
But when I really think about it, in a world where AI can do everything that humans can do, the only real edge any humans have is the fact that they are actually human. Professions such as art and therapy just happen to be some of the fields that rely on the inherent properties of being human. This analysis has definitely made me more open minded to the possibility of pursuing more creative or human centric professions for the future.
Of course with something as uncertain as AI, it is hard to say how the future will truly look. My analysis of these questions relating to human value and jobs after AGI is definitely in its infancy, so I am very open to hearing other perspectives. I am sure I might have missed some properties unique to humans as well as jobs that rely on them so please feel free to reach out with any thoughts!
