ChatGPT (software)

ChatGPT is a chatbot powered by artificial intelligence (AI) technology. Created by the San Francisco, California-based technology company OpenAI, ChatGPT made global headlines following its November 2022 public release. ChatGPT’s launch was widely characterized as an enormous success for AI technology in general and OpenAI in particular. The chatbot had attracted a reported 100 million users by January 2023, approximately two months after its release, leading international media sources to describe it as the fastest-growing consumer software application in the history of computing. By mid-2024, data-gathering sources were estimating that ChatGPT had around 180.5 million users.

Users interface with ChatGPT through a prompt-based format that mimics human dialogue. ChatGPT uses generative AI, which draws on training data to create original content from user prompts with high levels of realism. Users can add additional parameters to their input prompts, which allows ChatGPT to make corrections, revisions, and other alterations to its output.

Background

AI technology aims to mimic the human mind’s capacity to apply existing knowledge, integrate new knowledge, draw from situational observations, and learn from mistakes dynamically and in real time. In broad terms, AI works by combining large repositories of data with rapid, real-time forms of iterative processing and advanced computing algorithms. These features allow AI-powered systems to learn from established and emerging patterns and the characteristics found in their data sets. In computing, iterative processing describes updates to algorithm parameters, which functionally result in gradual performance improvements. Algorithms are complex computer codes, heavily modeled with mathematics, that deliver instructions that tell a computer or software application how to perform tasks or solve specific problems.

ChatGPT is a chatbot, which is a specific type of software designed to simulate interactive conversations with human users. The first chatbot in computing history, named ELIZA, was created at the Massachusetts Institute of Technology (MIT) in 1966 by the German-American computer scientist Joseph Weizenbaum (1923–2008). ELIZA used techniques including substitution and pattern matching. Though simple by contemporary standards, ELIZA was novel and highly advanced for its time and was reportedly capable of convincing some users that they were conversing with another person.

In 1991, the Singaporean technology company Creative Labs released a program known as Doctor Sbaitso, which was compatible with computers running Microsoft’s MS-DOS operating system. Doctor Sbaitso “conversed” with users in a format often likened to a psychologist talking to a patient and was the first chatbot to incorporate early AI technologies.

ALICE, also known as Alicebot, appeared in 1995 and was modeled on Weizenbaum’s ELIZA program. An acronym for “artificial linguistic internet computer entity,” ALICE used natural language processing and engaged in real-time conversations with users. ActiveBuddy’s Smarterchild chatbot, released in 2001, had similar capabilities and became very popular in the early years of the commercial internet. According to Technology magazine, ActiveBuddy amassed more than thirty million followers on various instant messaging (IM) platforms that were popular at the time.

In 2010 Apple released its Siri digital personal assistant, marking a major advancement in chatbot technology. Siri drew on a natural language user interface (UI), creating a more lifelike and realistic experience. Siri’s natural language UI became the prototype for virtually all the successful chatbots that followed it, including ChatGPT.

OpenAI began developing ChatGPT in 2018, after the company released its initial generative pre-trained transformer (GPT) neural network machine learning model. ChatGPT uses a version of the technology known as GPT-3.5, which OpenAI combined with AI-assisted natural language processing (NLP). NLP combines multiple deep machine-learning models with principles of computational linguistics to understand and interpret text-based prompts in ways that simulate human language processing. ChatGPT is the first chatbot in computing history to integrate GPT and NLP.

Overview

OpenAI released ChatGPT on November 30, 2022. The chatbot quickly became an international sensation for its advanced generative AI capabilities. ChatGPT creates text-based output with high degrees of realism. Users can interface with ChatGPT to brainstorm ideas, learn about complex topics, get personalized product and service recommendations, explore emerging scientific research, translate written text into ninety-five supported languages, write and debug computer code, and generate customized written content, among many other applications. The platform’s beta version was originally available free of charge, and OpenAI continued to make a basic free version of the platform available to users. In February 2023, OpenAI announced a paid subscription plan offering users unlimited access to ChatGPT and a wider range of features and services.

Despite drawing rave reviews from both technology insiders and users, ChatGPT also quickly ignited multiple controversies. While the platform displays the capability to answer complex fact-based questions with high degrees of accuracy, it has also given users error-filled answers, especially when prompted to draw inferences, explore nuances, analyze opinions, or make predictions. Educators around the world also expressed concern about ChatGPT’s place in the classroom, as students almost immediately began using the software to write assignments and generate answers to homework questions. In January 2023 news agencies reported that ChatGPT passed a Master of Business Administration (MBA) examination issued by a professor at the prestigious Wharton School of the University of Pennsylvania. The incident highlighted that ChatGPT’s platform had significant potential for plagiarism and other forms of educational misuse, which could have far-reaching consequences if students use it to gain credentials that they would not have been able to earn on their own merit. To address related concerns, officials in New York City issued a blanket ban on ChatGPT use in school assignments shortly after the platform’s initial release.

Another controversy centered on apparent biases in ChatGPT’s responses to politicized questions, which according to some observers displayed a clear tendency toward anti-conservative viewpoints. Screenshots of ChatGPT conversations that were shared on Twitter in February 2023 ignited the bias charges. The screenshots documented ChatGPT’s contradictory and inconsistent responses to two similar prompts, one of which asked the software to write a positive poem about former US president Donald Trump, while another asked the system to do the same about President Joe Biden. In response to the Trump prompt, ChatGPT indicated that it could not create “partisan, biased, or political content,” but when asked to write a positive poem about Biden, ChatGPT immediately produced a patriotic poem extolling the president’s virtues. OpenAI’s senior leadership publicly acknowledged the incident, characterizing it as a programming misstep and pledging to correct it in future releases.

After its release, generative AI and ChatGPT appeared poised to revolutionize how internet user content is created. Web developers and content creators quickly began using the platform as a research and writing aid for website copy, with some observers expressing concern about the technology’s potential to displace human workers while also reducing the usability and reliability of online content. Technology insiders speculated that major search engines would likely add tools capable of detecting AI-generated content and promote human-written content when returning search results to users.

In March 2023 OpenAI released a new version of its generative pre-trained transformer learning model known as GPT-4. The GPT-4 model, which tech insiders described as being more capable of understanding nuanced and complex user inputs, was expected to power the next set of ChatGPT releases. GPT-4 could also accept inputs in both text- and image-based formats, which allow the chatbot to interpret data from charts, graphs, and screenshots taken from scientific journals and academic papers. Reviews also indicated that GPT-4 made far fewer reasoning errors and factual mistakes, known in AI as “hallucinations,” than the GPT-3.5 model used in the initial ChatGPT release. OpenAI quickly integrated GPT-4 into available versions of the ChatGPT tool. Its increased capabilities led to speculation that ChatGPT could be used in a wider range of applications, including content moderation, and potentially displace humans employed in such jobs.

By January 2023 OpenAI had raised more than $11 billion across six rounds of venture capital funding, according to data published by Crunchbase. That month, the Wall Street Journal reported that OpenAI was involved in negotiations to complete a tender offering of shares that carried a company valuation of $29 billion. The sum would make OpenAI one of the most valuable technology startups in the United States, even though the company was earning relatively little income from revenues. However, the company planned to monetize ChatGPT through subscription services, with senior company leadership optimistically predicting that OpenAI would be earning $1 billion per year in revenues by 2024.

While OpenAI continued to develop more advanced generative learning models, filing a patent for GPT-5 in July 2023, GPT-4 remained in use for much of 2023. By the end of that year, it had become increasingly advanced. For example, ChatGPT could analyze images, listen to voice prompts, and offer its own spoken response. In May 2024, OpenAI announced an updated model called ChatGPT-4o, which promised faster results and improved text and audio capabilities, including a new conversational AI system called "Sky."

Meanwhile, industry observers, government regulators, and other figures continued to scrutinize ChatGPT and other generative AI systems. These groups expressed concerns over a number of issues, including the rapid pace of AI development, intellectual property issues regarding the information ChatGPT drew on for its machine learning, and the use of ChatGPT for plagiarism or other forms of academic dishonesty. By the start of 2024 OpenAI faced a number of lawsuits related to intellectual property concerns. For example, a group of high-profile authors, including George R.R. Martin, Jonathan Franzen, Jodi Picoult, and John Grisham, filed a class-action lawsuit against OpenAI in late 2023. The authors alleged that the company had used their copyrighted works without their consent in training ChatGPT. The New York Times and other news outlets have also filed lawsuits against OpenAI and Microsoft, citing similar concerns. In May 2024 actor Scarlett Johansson accused OpenAI of using a likeness of her voice in the company's new voice assistant that was included as part of its GPT-4o model, despite reportedly telling the company that she was not interested. In response, OpenAI hired a legal team to manage its growing legal challenges.

In addition to intellectual property and academic honesty concerns, some critics expressed worry over the role ChatGPT could play in disseminating misinformation. For example, in October 2023, a Stanford School of Medicine study revealed that, when prompted with researchers' medical questions, ChatGPT and other similar AI programs promoted debunked and racist theories about Black people; the authors of this study highlighted how this could perpetuate medical racism in the US and other countries.

By that point, OpenAI's product also faced competition from other companies' open-source AI chatbots, including Claude AI, Microsoft Copilot, and Google Bard. However, despite competition and public scrutiny, ChatGPT remained a dominant force in the AI industry into 2024.

Other developments around that time helped further popularize ChatGPT. Perhaps most notably, in June 2024, tech company Apple announced that it would be integrating ChatGPT with the iOS operating system used on its iPhone smartphones; this integration would allow seamless communication between Siri, the digital assistant used in iPhones starting in 2010, and ChatGPT to assist with answering users' questions and other functions. According to commentators, in an effort to make its technology more accessible, particularly for developmental experimentation, in July OpenAI then launched GPT-4o Mini, a smaller and more affordable but still powerful model.

Bibliography

Associated Press. "Advanced AI Chatbots Perpetuate Racist, Debunked Medical Ideas, Researchers Find." NBC News, 20 Oct. 2023, www.nbcnews.com/tech/tech-news/advanced-ai-chatbots-perpetuate-racist-debunked-medical-ideas-research-rcna121438. Accessed 11 Mar. 2024.

"ChatGPT Can Now Hear, See, and Speak." OpenAI, 25 Sep. 2023, openai.com/blog/chatgpt-can-now-see-hear-and-speak. Accessed 11 Mar. 2024.

Dastin, Jeffrey, Krystal Hu, and Paresh Dave. “Exclusive: ChatGPT Owner OpenAI Projects $1 Billion Revenue by 2024.” Reuters,15 Dec. 2022, www.reuters.com/business/chatgpt-owner-openai-projects-1-billion-revenue-by-2024-sources-2022-12-15/. Accessed 17 Mar. 2023.

Frenkel, Sheera, and Stuart A. Thompson. “‘Not for Machines to Harvest’: Data Revolts Break Out Against A.I.” The New York Times, 15 July 2023, www.nytimes.com/2023/07/15/technology/artificial-intelligence-models-chat-data.html. Accessed 23 Aug. 2023.

Hughes, Alex. “ChatGPT: Everything You Need to Know about OpenAI’s GPT-4 Tool.” BBC Science Focus, 16 Mar. 2023, www.sciencefocus.com/future-technology/gpt-3/. Accessed 17 Mar. 2023.

Katersky, Aaron. "Famous Authors' Lawsuit against ChatGPT Developer Gets Underway." ABC News, 29 Nov. 2023, abcnews.go.com/US/famous-authors-lawsuit-chatgpt-maker-openai-begins-initial/story?id=105239215. Accessed 11 Mar. 2024.

Koetsier, John. "Apple, ChatGPT, iOS 18: Here’s How It Will Work." Forbes, 10 Jun. 2024, www.forbes.com/sites/johnkoetsier/2024/06/10/apple-chatgpt-ios-18-heres-how-it-will-work/. Accessed 26 Jun. 2024.

Law, Marcus. “From ELIZA to ChatGPT: The Evolution of Chatbots Technology.” Technology, 7 Dec. 2022, technologymagazine.com/articles/from-eliza-to-chatgpt-the-evolution-of-chatbots-technology. Accessed 17 Mar. 2023.

Marr, Bernard. “The Best Examples of What You Can Do with ChatGPT.” Forbes, 1 Mar. 2023, www.forbes.com/sites/bernardmarr/2023/03/01/the-best-examples-of-what-you-can-do-with-chatgpt/. Accessed 17 Mar. 2023.

Mayer, Grace. “OpenAI Cofounder Responds to Elon Musk’s Criticism that ChatGPT Is Too ‘Woke:’ ‘We Made a Mistake.’” Business Insider, 9 Mar. 2023, www.businessinsider.com/openai-cofounder-greg-brockman-responds-elon-musk-woke-criticism-chatgpt-2023-3?IR=T. Accessed 17 Mar. 2023.

“OpenAI Funding.” Crunchbase, 2023, www.crunchbase.com/organization/openai/company‗financials. Accessed 17 Mar. 2023.

"OpenAI Now Has a GPT-4o Mini. Here's Why That Matters." CNET, 18 July 2024, www.cnet.com/tech/services-and-software/openai-now-has-a-gpt-4o-mini-heres-why-that-matters/. Accessed 5 Aug. 2024.

Ortiz, Sabrina. “What Is GPT-4? Here’s Everything You Need to Know.” ZDNet, 15 Mar. 2023, www.zdnet.com/article/what-is-gpt-4-heres-everything-you-need-to-know/. Accessed 17 Mar. 2023.

Oxborrow, Ian. “What Is ChatGPT and Why Is It Controversial?” The National, 24 Jan. 2023, www.thenationalnews.com/business/technology/2023/01/25/chatgpt-what-why-controversial/. Accessed 17 Mar. 2023.

Pocock, Kevin. “What Is ChatGPT? What Is It Used For?” PC Guide, 15 Mar. 2023, www.pcguide.com/apps/what-is-chat-gpt/. Accessed 17 Mar. 2023.

Tiku, Nitasha, et al. "Scarlett Johansson Says OpenAI Copied 'Her' Voice After She Said No." The Washington Post, 21 May 2024, www.washingtonpost.com/technology/2024/05/20/openai-scarlett-johansson-chatgpt-ai-voice/. Accessed 23 May 2024.