AI Sentience

AI Sentience is becoming a more and more popular topic in the coming future. People are talking more about Artificial Intelligence, and pondering the question of whether these AI can or cannot be sentient. The idea of sentience is derived from the modern thought of whether humans can generate a piece of machinery to think like a human. That is the basis of AI sentience, and the debate is over whether such a fantasy can happen for real or not. AI Sentience cannot happen, at least shortly, simply because the process behind machine learning, does not result in sentience, but instead creates AI competence.

What is AI?

AI simply put it, is machines intaking input given by the interface or the user, and using their derived logic and reasoning, to push out an output. The hidden networking and logic that is between the input and the output are unknown. The programmer nor the machine knows what it is doing during the time between the input and the output. Why is this? To understand this hidden mess, we must know how machines, not AI, are built. Programmers build machines by inputting a code that the programmer derived, and the machine will follow (Metz C 2022). In this instance, the machine is not thinking for itself but instead following the rules set by the programmer. When building AI. the programmer cannot just give a set of instructions for the machine to follow, then it won't derive its logic to perform tasks (Metz C 2022).

Instead, the programmer builds 2 types of programs: A building program to build student bots, and a teaching program to test student bots (Youtube 2017). This is the AI-building process. For example, let's say the coder wants the bot to tell the difference between 1 and 2. Easy for humans because we have neural connections in our mind that help us tell the difference between 1 and 2, but to a machine, those 2 inputs are just random source code images of 1s and 0's, with no real significance for the machine. Give a test to a robot, and it will fail badly, so the AI building process begins. The building bot starts creating student bots using random connections and interactions. These student bots like our first bot, cannot tell the difference between 1 and 2, there is no logical setup yet. That's where the teaching bot comes in to try to teach students between 1 and 2. The teaching bot has no idea how to tell the difference but instead receives a test from the programmer with the answers (Youtube). The teaching bots test the student bots with an exam and tell them to tell the difference between 1 and 2. These student bots who have no logical connections for these images will fail, by clicking random answers.

Most fail, but some through the random chance pass. Bots who failed to get recycled, but bots who pass go back to the building bot to get tweaked and rebuilt. The building bot sees the connections the passing student bots made and use that to redesign the student bots. Repeat this process again and again, for millions of student programs, with exams with millions of questions. After multiple interactions, tests, and connections were made, a student bot emerges where it can kind of tell the difference between 1 and 2 without random chance involved. That bot gets tweaked millions of times going through the same process until it can finally tell the difference between 1 and 2 without any external help or random chance. It had learned a key difference between those 2 images. Give any set of images ranging from 1 and 2, and it can tell the difference. Now open up this bot and the logical program that it developed is unknown. The programmer, the builder bot, the teaching bot, or this perfect student bot know what logical interactions were made to spit out the right output. The intelligence in AI is unknown by the programmer and the AI itself (Quility-Harper 2022).

Can AI be Sentient?

The idea that AI, can think for itself, cannot happen shortly. Though the process of machine learning showed that AI has a node-to-node connection pathway between a given input and their output, meaning they do think for themselves, the case is not exactly true (Metz R 2022). The AI itself does not understand how the connective pathway works. How does this mean that AI can think for itself? The problem is humanity is looking at this problem from the wrong perspective. AI being sentient is still a mystery, but AI is competent (Metz C 2022). Look at any AI system that was developed recently, going through the machine learning process, we can see the connections made by any recent AI program is unknown. The developer nor the program knows what that hidden connection is or how it works. That is the major difference between machine learning and human learning. Humans learn through emotional connections, and their neural pathways are developed through experience and memories. A human knows what he or she is learning at the moment, and remembers these neural connections they made like a machine, but instead of not understanding what connections were made, humans instead build more of their understood neural connections, generating sentience. An AI or a smart language program simply cannot do that because it has no understanding of the connections pathway it made to do its assigned job (Metz R 2022). It cannot register what it "learned" at the moment, instead like any machine, just use this new connection to perform its task. AI sentience is not developed yet in any of these new "AI sentient technology." Instead, those types of AI are competent, not sentient. Competence in the sense they can perform efficiently using their derived connections to give the developer, interface, or the user what they want. Instead of trying to find AI to be sentient, we must see their advancements when it comes to being competent. AI is getting better and better at what they are designed to do, which is to make a task easier and more efficient for the developer, interface, or user. That is the basis of machines, and AI fit that bill. Sentience can happen in the distant future if an AI can set up its machine learning process, and code itself without the need of a developer, sentience may be achieved, but as of now, AI is not sentient, they are amazingly competent.

Metz, C. (2022, August 5). A.I. is not sentient. why do people say it is? The New York Times. Retrieved September 3, 2022, from https://www.nytimes.com/2022/08/05/technology/ai-sentient-google.html

Metz, R. (2022, June 14). No, Google's AI is not sentient | CNN business. CNN. Retrieved September 3, 2022, from https://www.cnn.com/2022/06/13/tech/google-ai-not-sentient/index.html

Quilty-Harper, C. (2022, June 17). Can ai gain sentience? here's how google chatbot lamda answers human questions. Bloomberg.com. Retrieved September 3, 2022, from https://www.bloomberg.com/news/articles/2022-06-17/can-ai-gain-sentience-maybe-but-probably-not-yet-quicktake

YouTube. (2017). How Machines Learn. YouTube. Retrieved September 3, 2022, from https://www.youtube.com/watch?v=R9OHn5ZF4Uo.