Challenges ChatGPT, the bot used in academic work

Considered The biggest revolution technological Since the invention of electric lightGenerative AI is gaining popularity throughout 2022. The main one is chatfrom the OpenAI Research Center, which reached the public in December.

ChatGPT looks like a messaging app. In it, you can give any kind of instruction to the bot, which will analyze billions of pages, and generate a 100% original text answer. Among the uses are creating lists, slogans, descriptions of scenes from fiction, and articles on any topic.

This last app Teachers worried around the world. Because ChatGPT is a powerful model that works with billions of parameters and can discuss any topic available on the Internet, it has been used by primary and higher education students as a shorthand for academic work not being done.

In this text, the file the Nexus It explains what generative AI systems are, how they work, what concerns they raise, and how these problems have been dealt with.

How does ChatGPT work?

ChatGPT, from OpenAI, works based on an algorithm called GPT-3, which is a natural language processing paradigm – a set of technologies and principles that allow a machine to communicate with a human in a natural way. This is the case with voice assistants Alexa, from Amazon, and Siri, from Apple.

In addition to voice assistants, natural language processing technologies can be used to generate texts. In 2022, the the Nexus First published a script written in part by GPT-3.

GPT-3 works with a vocabulary of thousands of words, knowledge of grammar, and the ability to create hundreds of billions of different sentences.

Upon receipt of the topic, the bot analyzes the suggested content and writing style. From there, generate the text. The more detail in the proposal, the better GPT-3 will be at generating similar text.

The work of a bot like GPT-3 can be impressive in terms of results, but it is far from comparable to human cognition when it comes to shaping thinking. This is because the algorithm knows, for example, how to form sentences, but it doesn’t understand exactly why sentences are built that way. What may seem beautiful poetry to the reader is interpreted as just another series of instructions by the machine.

In 1950, the English mathematician Alan Turing developed the Turing Test – or the “Imitation Game”. It consists of a series of questions asked to determine whether a robot can think indistinguishably from a human. Even today, No algorithm I managed to pass the challenge. However, it can be difficult (or almost impossible) to detect that a text was written by a bot and not by a person.

Challenges in education and science

Basic and higher education students have begun to use ChatGPT in school work as a “short cut” for not doing tasks suggested by teachers.

In New York, public schools Banning the use of ChatGPT On devices connected to educational units networks. “Due to concerns about negative impacts on student learning and concerns about content security and accuracy, access to ChatGPT is restricted on New York City public school networks and devices,” said a statement from the city’s Department of Education at the beginning of 2023. .

At Northern Michigan University, Philosophy major used ChatGPT to write what was considered “Best work in the class” by Professor Anthony Aumann. After that, the app is blocked in the organization. There are no public records of anything similar happening in Brazil yet.

The ChatGPT problem seems to be an advanced version of what happened in the early 2000s, when students of all levels began copying and pasting ready-made texts from the Internet for their assignments. The difference is that ChatGPT generates unique text, which technically cannot be considered plagiarism.

In scientific production, ChatGPT has been used to write an introduction to an article published in the bioRxiv repository. The bot-generated text was not detected by the impersonation identifiers nor by the scholars who reviewed it. In a text on the Nature blog, the scientists were unable to pinpoint Moral implications This type of technology applications.

Differences about usage

Kevin Rose, a technology columnist for The New York Times, said he believes teachers should not fight technology, but use in the teaching process.

“ChatGPT can be a powerful ally for the educator. The answers given by the bot can be a first step for students to develop critical sense, learn to verify information and teach computing principles,” he said in the Jan. 12 text.

Roose also argues that generative AI systems such as ChatGPT can be useful in making teachers’ lives easier when preparing lessons, quickly creating instructional plans and exercises based on parameters given by teachers.

Daniel Hermann is a high school teacher from California He was less optimistic In an article he wrote for The Atlantic magazine. “Everything is made, in fact. School articles. Invented. Grammar? Invented. Writing itself. Invented. As of now, OpenAI makes us wonder if these things are worth preserving within a school setting.”

“We are in a moment like the invention of the calculator, which saves me a lot of time doing long division, or more like the invention of the pianist, and as a result, [o ChatGPT está] We have been robbed of what can only be communicated through human emotion? “

technology revolution

Even without a human-like thought process, generating AI like ChatGPT has amazing results. For this reason, experts considered it a revolution.

The World Economic Forum predicts that this industry can Generating “Trillions of Dollars Annually” In the near future, by allowing automation of certain repetitive, trivial or error-prone tasks – such as creating work schedules in companies or creating weekly cost spreadsheets.

Generative artificial intelligence is a Deep technological revolutionand will impact the most diverse industries and potentially help some of the world’s most complex issues, such as health systems and climate change,” Sylvain Duranton, technology columnist for Forbes, said Jan. 7.