Implementations, Morals, and Ethics
by: Nic Cheng
​
A simple breakdown of what it is, what it does, and why it matters.
The future is now.
It's ours to define.
Or not...
With rapid advances in the innovation of Artificial Intelligence, it may not even be our decision to make.
This article could have been written by Artificial Intelligence.
Only that it isn't. That's academically dishonest. But how would you know?
​
Well, you would open up a website that detects this behavior. Word by word, it would analyze patterns to see if it is more likely that a human wrote it than a robot. But that too is an artificial intelligence-powered program. Scientists must tread carefully when creating new AI. It's why they need to take ethics classes before they go out into the world and innovate!
One day, we may have our own personal AI assistants! Wouldn't it be lovely to sleep in while it takes notes in class for you?
Wait a minute. Full stop. What is Artificial Intelligence?
Artificial Intelligence (AI) is the ability of a computer system to think and act like a human. There are a wide range tasks that can be achieved by AI, such as descriptive, predictive, and prescriptive reasoning. At its core, the nature of AI will allow it to meet/surpass the bounds of human knowledge and creativity. What makes AI particularly powerful is its ability to efficiently parse through data and recursively train itself to recognize patterns. These processes are optimized to the point where scientists have trouble tracing how an AI achieved their output, creating what's known as a black box (Bagchi). Being unable to trace how they arrive at their output, scientists are unable to find out why an AI is accurate (or why it isn't-- more on that later) (Bailey).
​
There are three categories in which we categorize all AI, whether it's ones created in the past or ones we have yet to create.
1. Artificial Narrow Intelligence
The lowest-level form of AI. Technological and research bottlenecks restrict scientists from being able to break past this part of AI. Even our most powerful machines and current research allow for any of our AI programs to be classified under AGI. These forms of AI are trained to be able to accomplish a singular task. They are not considered sentient and are unable to think "freely."
​
2. Artificial General Intelligence
The level of Artificial Intelligence that is classified to be able to perform at the same level as the average human brain. This means that an AI must be sentient (can think "freely") and is able to learn and retain information. Through our current research, this could only result from a large enough neural network with enough processing power.
​
3. Artificial Super Intelligence
The peak level of Artificial Intelligence-- due to our technological and knowledge restrictions, this type of AI is completely restricted to science fiction (for now). Scientists and theorists have various predictions on how an ASI would function/interact with humans, but all of this is purely speculation with no formal data to back it up.
​
​
​
​
Extra Key Concepts
​
Some readings and other sources may use some extra terminology. Here are a few that are good to understand before diving deeper:
​
Machine Learning
The most basic form of ANI, Machine Learning programs are extremely advanced tools of statistical analysis. Using information gathered from a training data set, patterns are randomly stored to enable an ML program to predict an outcome.
​
Neural Networks
A computational/mathematical representation of the human brain on a computer. This is used as a shell to host the learning/processing environment of an AI in both the development and production stages.
​
Deep Learning
The process of training an Artificial Intelligence program similarly to the way a human would learn/memorize a subject.
Alright. So where can we find Artificial Intelligence?
Artificial Intelligence is already very common across our digital society. Many aspects of our lives are already being shaped and reshaped by AI!
In terms of AI in society, they can be divided into just two categories: AI for private use and AI for commercial use. Private-use AI can be used by everyday internet users, like us. Whether it be to answer an email or just to play around with, they change a small part of our lives. Commercial-use AI, on the other hand, is large-scale and everywhere. Corporations utilize AI across their infrastructure to optimize routines and minimize overhead costs.
​
Here are some examples of private-use AI:
-
ChatGPT (its website, for personal use) - Used to generate text output and some forms of data visualization.
-
Midjourney, Stable Diffusion (its website, for personal use) - Used to generate original images.
-
Tesla FSD (Full-Self Driving) (in its cars) - Used to autonomously operate a car without human intervention.
-
Microsoft CoPilot - Used in the workplace, aids individuals to complete day-to-day office work (like writing emails, creating schedules)
-
Codex - Used by programmers to generate segments of codes, or even entire programs.
​
Here are some examples of commercial-use AI:
-
ChatGPT (its API): used by corporations to replace human tech support and human customer service operations.
-
Midjourney, Stable Diffusion (its API): used by corporations to generate images and visualizations.
-
Various AI-powered self-driving programs: used to automate taxi services in certain parts of the United States. Currently in beta, and clashes with lawmakers.
-
Blockchain and Supply Chain: used by large distribution companies (ex. Amazon) to optimize shipping routes to minimize delivery times to customers and lower operational costs/overhead.
-
Advertisements: companies like Target and Google utilize AI to predict the products that users from certain demographics may want to purchase and send them targeted ads (Anne).
So AI will just benefit humanity, right?... Right??
Yes, but not quite. When a product, not just with AI, is created, it's made for a set of intended consequences. This means that the product was created to serve a certain purpose, as envisioned by its creator. However, there are also unforseen unintended consequences, which are effects of a product that were not intended to happen. Take ChatGPT, for example: it was created to be a display of a powerful machine learning language model, but is also used by students to cheat in school assignments. Its ability to generate human-sounding text has also made it a tool for foreign adversaries to spread disinformation on social media (Fung). Countries such as Russia are currently in the process of disseminating such disinformation ahead of the 2024 election (Schneier). In addition, ChatGPT is also able make predictions to aid in stock market trading (Jain).
​
In the video below, renowned Oxford philosopher Nick Bostrom explains how some of these unintended consequences can occur at a catastrophic scale in future Artificial Super Intelligences.
​
Lastly: Should we be doing all of this?
Artificial Intelligence has arrived and it's here to stay. With all the benefits that AI boasts and all of its untapped potential that it boasts, it's tempting to implement it wherever possible. However, it's important to not forget that the speeds at which AI is developing makes it difficult to create a foolproof cybersecurity infrastructure and for lawmakers to pass laws to limit the use of AI for negative purposes. Even though most large industries have already adopted AI within their everyday processes, over 50% of CXOs say that AI poses a large security risk within their company (Sharma). In addition, these same industries use private user data to train their AIs-- this combination of user data in the open and lack of proper cybersecurity should be alarming to the general public.
Such an oversight and lack of restricted use of AI in both federal and commercial capacities should sound as an alarm to society as a whole. As discussed before, the vast benefits that AI can be used to serve society is great, but by granting AI increased control also comes increased risk. As stated by Nick Bostrom, society needs to be prepared to contain AI when it becomes powerful. Lawmakers should restrict the use and development of AI until it can be properly regulated and controlled, while also allowing society to innovate and benefit from AI.
​
"MACHINE INTELLIGENCE IS THE LAST INVENTION THAT HUMANITY WILL EVER NEED TO MAKE."
- Nick Bostrom
So what do you think?
Whether we like it or not, AI will have constant impact through the rest of our lives. We can choose to embrace it, reject it, or find some happy medium in between. Regardless, society must tread carefully in its development and always be obligated to do what's morally and ethically correct.