AI Sentience and The Problems that Arise With It

Neeraj Gogate

Allen STEAM Center

Computer Science III

David Ben-Yaakov

September 2, 2022

The current media landscape, most often science fiction pieces, depict Sentient Artificial Intelligence (AI), the development of machines that simulate human actions, as an apocalyptic future that will doom the human race. This can be seen through the movie Avengers: Age of Ultron, which depicts the titular villain, Ultron, as he immediately tries to eliminate the human race after reading all of human records. While this type of story is perfect for works of fiction through its thrilling plot, this type of idea does not track very well to real life, as Sentient Artificial Intelligence would likely result in them integrating into society very well.

AIs Today

In its current form, AI is truly nowhere near sentient. Many AIs today are made for much simpler tasks and still have imperfections in the way they’re created (Metz 2022). Most AI’s today boil down to a combination of parsing through data looking for relevant information and an algorithm based on past statistics collected by data scientists (Burns, 2021). An example of one of these algorithms are algorithms used by either YouTube and TikTok, which use their algorithms to figure out what videos users are watching and interacting with the most, and then recommending videos that match the user’s tastes. This algorithm is incredibly effective and accurate at predicting a user's tastes after the user watches about 50-60 videos, but there are some problems with the algorithm, specifically that it’s too good at what it does. Often the algorithm gets stuck in a niche that's impossible to get out without watching a large amount of videos. This sounds great at first, but this is problematic with the high abundance of political videos on YouTube. Getting stuck in a niche in political style videos often happens at extremely polarizing parts of the political spectrum, and getting stuck at these ends often means that people can’t get to the opposing end of YouTube which means that people aren’t exposed to as many differing views and can become stuck in their ways. A newer form of AI has gotten a bit more complicated with images parsing, with DALL-E. DALL-E takes images from the internet and uses them to generate images from user entered prompts (Roose, 2022). However, it does have some problems as it can produce some images which aren’t entirely perfect. While it is rather great at making landscapes and environments, it isn’t as great at making characters as it can often smudge in certain cases. Despite these problems, however, these AIs are constantly improving, and eventually we may reach a point where these AIs are virtually indistinguishable from humans.

AIs Won't Revolt

If AIs were highly specialized machines meant to complete a large variety of tasks done by humans rather than the highly specialized AIs with no proficiency in any other skills the world is trending towards, then the AIs would most likely be on the side of humanity rather than being the ones to cause an apocalypse .The primary reason for this is that they would most likely reinforce ideals of unity and peace present in our world today. If the AIs were to parse through and read through data present on the modern internet through reliable sources, they would most likely pick up ideals that are similar to the ideals held in countries with reliable internet connections, as those would be the ones most commonly held on the internet. This would result in the AIs having an ideology of the average of all internet users, thus making them have rather agreeable positions that would not conflict too heavily(unless there is a huge disparity in opinion). This would allow the AIs to be nothing more than an assistant to humanity, as they would allow humanity to complete tasks at a much faster rate than humanly possible, while also solving problems and allowing humanity to focus on smaller issues. Solving problems such as food shortages, water problems, and housing inequality across the globe with these AIs through removal of most of the middle management by humans would result in humanity being able to focus on issues that are rather trivial comparatively. Since people’s basic needs will be met relatively easily, the need for capitalistic systems will no longer apply. People would be able to focus more on what they are truly interested in as the need for advancement will be depleted significantly. This would result in less competition among people, drastically removing a lot of the mental health issues people have, as people are able to be more relaxed in their jobs. It would essentially create a utopia for humans. Despite this, there is one critical flaw with this type of technology.

Ethics of Using AIs

While the AI being sentient would most likely not cause many problems relating to the extinction or enslavement of the human race, there would be an extreme problem regarding the ethics of letting the robot complete tasks. This problem would be related to the definition of sentient being “responsive to or conscious of sense impressions” (Merriam-Webster, 2009). If an AI was truly able to have emotions (for the sake of argument, let’s say through some chemical engineering relating to brain wave patterns) utilizing it for working rather than humans would bring into question whether they have certain rights. If they are able to simulate human emotions and actually in some way feel human emotion, then they would need to have some sort of protection in the law, as treating them as inferior to humans would be deprecating to them and border on the lines of enslavement.

Although a form of highly-advanced sentient AI can be extremely beneficial to society, there are major problems relating to the morality of letting them be sentient and treating them as tools to society rather than sentient beings.

References

Burns, E. (2021, March 30). What is machine learning and why is it important? SearchEnterpriseAI. Retrieved September 2, 2022, from https://www.techtarget.com/searchenterpriseai/definition/machine-learning-ML#:~:text=Machine%20learning%20(ML)%20is%20a,to%20predict%20new%20output%20values.

Metz, C. (2022, August 5). A.I. is not sentient. why do people say it is? The New York Times. Retrieved September 2, 2022, from https://www.nytimes.com/2022/08/05/technology/ai-sentient-google.html

Roose, K. (2022, August 24). We need to talk about how good A.I. is getting. The New York Times. Retrieved September 2, 2022, from https://www.nytimes.com/2022/08/24/technology/ai-technology-progress.html

sentience. In Merriam-Webster.com. Retrieved September 2, 2022, from https://www.merriam-webster.com/dictionary/sentience