Has ChatGPT ever not been in the news since its launch in November 2022? It's created excitement and panic. On June 26, it was announced that US Congress members cannot use the regular ChatGPT—they must use the paid version, ChatGPT Plus, purportedly for privacy reasons. There are also mixed messages regarding regulatory concerns in the European Union. The week prior, a study deemed that ChatGPT is inherently in violation of EU law.
This chatbot, which has both a free and paid version, launched by the research company OpenAI is one of the most prominent recent examples of artificial intelligence (AI).
In only a few months of existence, ChatGPT has quickly dominated conversations about AI and risen to international prominence. Users have adapted ChatGPT for a wide variety of uses, including to create workout plans, craft bedtime stories for children, and even to write scripts for television shows. Most importantly, ChatGPT has brought AI technology into mainstream awareness, providing a vehicle for people new to AI to imagine the possible ways that humans can interact with machines. In this article we explore what ChatGPT is and how it became popular so quickly.
The Basics of ChatGPT
ChatGPT is an example of an AI system that understands and produces conversational dialogue. It is a type of AI tool known as a chatbot, which many consumers will likely have encountered on the customer service websites for a variety of companies. OpenAI, the developer of ChatGPT, is a tech research company that aims to ensure artificial intelligence benefits all of humanity.
The name ChatGPT contains a reference to “Generative Pre-training Transformer,” which indicates the way that Chat GPT processes language. A key element of the way that ChatGPT was trained—and something that sets it apart from chatbots over many years—is its use of reinforcement learning from human feedback (RLHF). RLHF is a process in which human AI trainers work with AI systems, providing rewards to help ChatGPT learn and correct its mistakes, to answer follow-up questions, and to automatically fix incorrect assumptions.
In preparing this article, Decrypt asked ChatGPT to “write a paragraph explaining what ChatGPT is and how it works.” In response, ChatGPT said:
ChatGPT is an advanced language model developed by OpenAI, built upon the GPT-3.5 architecture. It functions as a conversational AI, designed to engage in human-like text-based interactions. By leveraging deep learning techniques, ChatGPT has been trained on an extensive corpus of diverse internet text, encompassing a wide range of topics and styles. This training enables it to generate coherent and contextually appropriate responses based on the input it receives. ChatGPT operates by employing a transformer neural network, which allows it to process and understand the relationships between words, sentences, and concepts in the given text. It can infer meaning, recognize patterns, and generate relevant and coherent responses based on the context provided in the conversation. Additionally, OpenAI has implemented strategies to encourage safe and responsible usage of ChatGPT, including mitigating biases and providing user control over the system's behavior.
To further explain some of ChatGPT’s response, it is a system making use of deep learning, a process by which an AI tool examines a vast quantity of data to learn about common rules, style, and context. This, as well as the fact that ChatGPT utilizes a transformer neural network, allows it to not only understand the literal and contextual meanings of a wide variety of prompts, but also to create responses based on those prompts in a way that closely resembles human dialogue.
How ChatGPT Can Be Used
As mentioned above, users have already uncovered a tremendous variety of use cases for ChatGPT in its first months of existence. These uses run the gamut from minimal, chatbot-like functions to creative partners in a wide range of areas. ChatGPT can create content ranging from essays, think pieces, and articles to computer code, emails, and much more. The fact that ChatGPT remains available in a research-and-feedback-collection stage, meaning that it is free to the public to use, means that people everywhere can take advantage of its myriad uses. This is one reason why the Swiss bank UBS has called ChatGPT the fastest-growing app of all time, with 100 million active users just two months after its launch.
How to Access and Use Chat GPT
The easiest way to begin to access ChatGPT is to visit OpenAI’s website and follow instructions to create an account. After you sign in, you’ll immediately have access to ChatGPT and can begin a conversation by asking a question. While the system is still in a research phase, it is likely to remain free and available for all users.
There is also a subscription option for users wishing to guarantee access even while the platform is at capacity, or to connect ChatGPT to other tools through plugins. And OpenAI is launching ChatGPT apps, web browsing features, and more.
What Are Some Controversies Surrounding ChatGPT?
A major concern surrounding ChatGPT is safety and security. The tool is undeniably powerful, and it has led to rampant speculation about what might happen when bad actors make use of ChatGPT. Could terrorists use ChatGPT to falsify documents, or to hack into otherwise-secure networks? Could ChatGPT be put to work creating code that could be used for nefarious purposes? The system is built with many security protocols in place, but the huge range of potential uses may leave vulnerabilities.
Another concern regarding ChatGPT is inherent bias. Because ChatGPT has trained on human-generated documents and data, and because humans have implicit and explicit biases, OpenAI has said that its systems are prone to bias on at least racial and gender lines. The more popular ChatGPT becomes as a tool, the more these biases may feed back into the AI-generated content that humans read and use. Biases can also contribute to damaging and inaccurate responses.
Unsurprisingly, one of the early controversies regarding ChatGPT has been its potential use by students looking to cheat at school. ChatGPT recently proved that it can pass the bar exam, and universities have raced to put in place tools designed to determine whether student work is created with the use of ChatGPT or a similar generative AI tool. But a bigger concern may be that students can interact with ChatGPT to cheat in ways that are more subtle than having the chatbot take an exam or write a paper. For instance, ChatGPT can easily produce a thorough outline for a paper without writing the essay itself. An enterprising student can use ChatGPT’s outline as the basis for a student-written paper that will easily pass through any detection system.
Although it’s not a controversy exactly, one limitation to ChatGPT is that it has been trained on data up to a certain time only. This means that ChatGPT may not be aware of the latest news and developments, potentially leading to inaccurate or incomplete responses to certain prompts.
ChatGPT does not typically ask for clarification when a prompt or question is ambiguous, and instead makes a guess about what the question means. This can lead to unintentionally incorrect responses. Stack Overflow moderators have said that a problem with the system is that it produces answers that “have a high rate of being incorrect,” while typically looking like they “might be good…and the answers are very easy to produce.” ChatGPT is, after all, putting words together in order that matches up with the data it has been trained on. This does not, however, necessarily mean that it understands a question or that it is answering correctly.
Possible Impact of ChatGPT
AI experts have speculated that ChatGPT could fundamentally transform the way that people interact with computers. ChatGPT may replace traditional search engines by providing users a more interactive, more responsive, and generally more accurate and functional search experience. This is perhaps one reason why Microsoft moved quickly after the release of the chatbot to launch a version of its Bing search engine that is powered by ChatGPT.
Though ChatGPT is not without its controversies, as illustrated above, many feel that its possible uses are seemingly endless. What’s more, OpenAI is continuing to develop the technology behind ChatGPT—and a host of other generative AI tools are also in development, pushing the AI space even farther forward.