September 11, 2024

Bowen Zhou

12th Grade

St. Francis Preparatory School 

Artificial intelligence, or AI, is becoming the next big thing – some of the biggest companies around the world like Google and Apple are pouring billions of dollars into it. It has also seen its surge in things other than summarizing or working out problems, it's also being used for art, music, and image generation among other things. 

Despite its amazing ability to create and solve problems, it also has disadvantages like cost, biases, unemployment, and privacy. Biases are a huge issue for AI, as it becomes more and more mainstream. It can expose people to the wrong information and could put people, who are none the wiser, in danger. For example, there was an incident when Stable Diffusion, a generative model that creates images through text prompts, thought the world was run by white male CEOs.

The data used to train AI can come from a variety of places like Reddit, which started requiring users to access its API in hopes of getting data and selling it to companies. It can also come from places like the government or companies that sell this data. They can also scrape data from the internet, but this comes with the risk of infringing on copyright. 

Bias can also come from the learning model used. But there are ways this can be stopped or reduced, for example, data can be diversified by using data from around the world and from a vast amount of sources. The implications of this are huge and the effort to stop AI bias is paramount. It can further harm groups who are already marginalized.

Another shortcoming of AI is the cost, not just the monetary cost but the environmental cost as well. This is a huge issue as artificial intelligence is getting used more and more by people every day. It is also important as we distance ourselves from fossil fuels and lean more toward the direction of the preservation of nature. For example, GPT-3, used as much electricity as 177 homes in the US for training the AI. The cost of AI will only get more intense as more and more powerful LLMs (Large Language Models) come out.