Jonathan Rueffer | Science Editor
ChatGPT and other artificial intelligence (AI) applications have become increasingly widespread and impactful, including in academic settings where ChatGPT’s conversational power can be used for studying, research and writing. The unprecedented scale and capabilities of ChatGPT make it increasingly important for people, especially students, to understand what it is and how it works.
A simple one-sentence definition of ChatGPT is that it is a generative AI chatbot. To be more specific, this means that ChatGPT is a conversational software product that lets users ‘chat’ with a large language model (LLM), which is a type of AI model trained on extremely large amounts of textual data to understand and generate human language. The ‘GPT’ in ChatGPT comes from the specific type of LLM used—a Generative Pre-trained Transformer (GPT) model. ChatGPT is considered an example of generative AI because its output is novel, i.e. the response has never been written anywhere before. During the training phase, the model ‘learns’ through intensive computations to determine numerical parameters that abstractly reflect human language and understanding. For example, the GPT-3 model has a total parameter count of 175 billion. When users prompt ChatGPT, the trained model can then make fast, real-time predictions using these learned parameter values.
ChatGPT was developed by OpenAI, a machine learning and AI startup founded by CEO Sam Altman, Greg Brockman, Elon Musk and others. Although it was initially created as a nonprofit, it now operates as a hybrid venture, having received substantial investment from Microsoft exceeding $13 billion since 2019, per IBM. OpenAI invented the GPT model, which other AI chatbots use, like Microsoft Copilot, the DALL-E image generation model and the text-to-video model Sora. Since OpenAI’s release of ChatGPT in 2022, its popularity has significantly influenced the current AI boom that has transformed major companies and industries. By January 2023, ChatGPT had become the fastest-growing software application in history, gaining more than 100 million users in two months, according to a report by The Guardian.
Today, alongside standard text generation, ChatGPT has been improved to generate images and hold audio conversations. ChatGPT is available for free online with an OpenAI account, but paid subscriptions can unlock more powerful models, including deep reasoning models and Sora. Other tiers are designed for the enterprise and education levels, with both options excluding personal input data from model training for additional privacy.
Some of the major limitations of ChatGPT include what is known as ‘hallucinations’, where the underlying algorithm identifies patterns in its training data that do not exist in real life, thereby generating a nonsensical output. Additionally, ChatGPT’s output can reflect biases present in the human-created content that it is trained on.
One of the major ethical concerns of ChatGPT is copyright infringement since GPT models have been trained on copyrighted content, which has led to OpenAI being sued by multiple publishers, including The New York Times and The Chicago Tribune. In response, OpenAI published an open letter in March 2025 that argued that the training should be allowed for OpenAI and other American AI developers to ensure U.S. global AI dominance.
Privacy is another concern as OpenAI collects user input data to further train its models, potentially compromising any submitted confidential or sensitive information. Moreover, ChatGPT is closed-source, meaning that third parties cannot examine its internal workings. Another concern is that ChatGPT can be used for plagiarism, misinformation and deepfakes.
Other ethical concerns focus on ChatGPT’s lasting environmental impacts. For example, a single query using 10 times as much electricity as a standard internet search. Additionally, a 2023 study by researchers at the University of California estimated that around 700,000 liters of freshwater were used in data centers to train the GPT-3 model.
With the increase in ChatGPT usage and overall AI boom in the U.S., it is important to be aware of how these transformative technologies work and operate, as well their limitations and related ethical concerns. At the Ohio State University, for example, an AI Fluency initiative is starting to be implemented, with a goal of having graduating students be fluent in both their major and artificial intelligence.