Artificial Intelligence, or AI, has dominated headlines and everyday conversations since the commercial release of ChatGPT on November 30, 2022. With the ability to write bodies of text, code computer programs, and create digital art, concerns have arisen over the prospects of AI replacing human jobs and creative projects. The reality of artificial intelligence is not so grim, and there are ways for everyone, even those without math degrees, to understand how it works.
The simplest way to picture the inner workings of an artificial intelligence network is as a mathematical function. Take, for example, the function y = mx + b. If you recall from algebra class, y is the output of this function, x is the input, and m and b are constant values which we can call parameters. In general, the artificial intelligence frameworks available to the public utilize neural networks. These are simply large functions which you can give input and receive output. For a natural language processing neural network such as ChatGPT, the words you give it are converted to numbers, plugged into the function, and the output is converted back into words.
Now I will move into a slightly more complex explanation. Although neural networks are simply large functions, it is key to understand that they have a mind-boggling amount of variables they take into consideration, not just a single one as in y = mx + b. Neural networks have multiple input (x1, x2, x3 . . .) and output variables (y1, y2, y3 . . . ), with multiple parameters corresponding to each input variable. For something like ChatGPT, you can imagine that each word in a given prompt corresponds to an input variable, and each word in the output corresponds to an output variable. This is how ChatGPT can answer multi-word prompts. Computer scientists train neural networks by feeding them billions of sample input values and then letting the network tweak its parameters so it gives the correct output for any given input. The method the network tweaks its parameters through is called backpropagation. There are generally two methods of training: supervised and unsupervised. For the former, computer scientists determine the correct outputs manually. For the latter, there is code in place that takes in raw data and figures out for itself what is correct. The latest version of ChatGPT, released in March 2023, has about 1.7 trillion parameters, demonstrating the scope of artificial intelligence frameworks.
Although AI may appear frightening, I see it as the next leap in the human journey through technology. Just as computer programs came to control machines instead of humans, AI can replace tasks requiring no creativity or human intuition. Just as jobs in the service sector increased to account for the rise of automation in the labor force, new jobs will appear in a world that has incorporated AI. How fast these jobs arise will come down to government policies and the shifting of the market.
One hope I have for AI is that it will democratize education in areas without access to quality educational resources. In rural areas of developing countries, many families cannot afford to send their children far away to secondary and above schooling. With the advent of groups such as The World Bank providing technology in rural areas combined with new AI systems that can provide tailored tutoring for students where teacher resources are scarce, children can receive higher levels of education. Higher levels of education are linked with lower levels of poverty, and this could be one step toward the United Nations’ goal of eliminating poverty in the world.
This concept of human distinctness emphasizes a key takeaway from understanding how neural networks operate: they can only predict patterns from previous data. As a human, you have the ability to innovate on new ideas and showcase your creativity. In this changing world, we all must discover what makes us special and distinctive from machines.
If you are anxious about artificial intelligence replacing the skills of writing and art, I urge you to find a way to make your pieces unique and distinct from AI. This is a skill that will become crucial in the coming technological epoch. Although artificial intelligence is capable of writing cohesive pieces of writing and making art, a computer can never be imaginative or make ideas of its own: it can only mimic the data it has trained on. As a human, you have the ability to innovate on new ideas and showcase your creativity. In this changing world, everyone must discover what makes them special from machines.