ChatGPT, launched on November 30, is an artificial intelligence (AI) chatbot developed by OpenAI. It utilises OpenAI’s GPT-3 family of large language models (LLM), a groundbreaking technology that uses insights gathered over a large corpus of data to accomplish a wide variety of tasks. ChatGPT has been fine-tuned using both supervised and reinforcement learning techniques to respond in a conversational way, resulting in a highly advanced natural language processing tool. With ChatGPT, users can engage in conversations that feel natural and human-like. Its wide range of potential applications includes providing assistance with writing emails, essays and code, as well as answering inquiries.

Since its release, ChatGPT has rapidly gained attention. Within five days, it had reached one million users. Due to its rapid surge in popularity, the ChatGPT has become a surprising sensation that worries even Google. Now, experts at Swiss bank, UBS, believe it to be the fastest-growing consumer app in history. According to a UBS analysis, ChatGPT had 100 million active users as of January, just two months after its inception in late November. The growth of ChatGPT has been much quicker than the nine months it took TikTok to achieve 100 million users and the two and a half years it took Instagram to do the same. Google and Facebook each took around five years in a far earlier era to reach 100 million users.

To start a conversation using ChatGPT, simply log in and ask a question. As ChatGPT is still in the research stage, it is currently free to use, and you can use it as many times as you like. This powerful AI bot has the potential to revolutionise the way we interact with technology and holds promise for the future of super-intelligent artificial intelligence. This model is capable of much more than just providing basic answers; it can write essays, describe art in great detail, generate AI art prompts, engage in philosophical discussions and write code for you.

The impressive capabilities of ChatGPT have generated a buzz, resulting in many people rushing to use it. This has resulted in new users being unable to sign on at most times of the day apart from the least busy. Since the website runs on a server, when too many users hit the server, it becomes overloaded and unable to handle requests. ChatGPT reaches capacity quite frequently. This does not mean that you will never get access to it. It just means that you should try accessing the website later when there are fewer people attempting to use it. Alternatively, you can leave the tab open and simply reload it every so often, keeping your session active.

If you want to go around the queue and enjoy dependable access, OpenAI has a choice for you. On February 1, OpenAI launched ChatGPT Plus, a pro membership that provides customers with broad access even during busy hours, quicker response times and priority access to new features and enhancements. The monthly fee for the premium service is $20.

ChatGPT also raises serious questions about the misuse of AI. Some users are attempting to make ChatGPT write malicious codes. Chatter analysis of Dark Web forums reveals that efforts to leverage OpenAI’s chatbot to assist with malware coding are already in progress. In the brief period that it has been available, it now appears that some have tried to use it to help write dangerous malware.

The chatbot signed up one million users within five days of its release. By January, ChatGPT had amassed 100 million monthly active users, only two months after its launch, making it the fastest-growing consumer application in history.

According to OpenAI, the language model was trained using reinforcement learning from human feedback (RLHF). Conversations in which humans played the roles of both the user and AI assistance were delivered to the model by human AI trainers. A separate AI model trained to detect harmful content and instruct ChatGPT to refuse to answer is also active, screening questions users submit. This makes it likely that ChatGPT refuses to complete a harmful prompt.

While ChatGPT appears impressive, it has its limitations. The purpose of ChatGPT, a language model, is to engage in discussions with users. To do that effectively, it has to be able to distinguish fact from falsehood. Search engines index web pages on the internet to assist users in finding the information they seek, but this option is not available in ChatGPT. This may lead to the model generating false, but correct-sounding information. As a result, StackOverflow, a developer question-and-answer website, has temporarily banned ChatGPT-generated answers to questions. “The main issue is that ChatGPT frequently generates incorrect responses, which is problematic,” according to a post by StackOverflow moderators. “I always urge individuals to verify any information they receive from me or any other source,” the bot sometimes adds, emphasising that its responses should not be treated as factual. OpenAI also acknowledges that ChatGPT occasionally provides “plausible-sounding yet inaccurate or nonsensical answers.”

While some people use ChatGPT for highly intricate tasks such as coding or creating harmful software, the chatbot can also be used for less complicated activities like engaging in polite conversation. However, there are concerns that AI chatbots may eclipse or impair human intelligence. For example, the chatbot could theoretically replace the need for a human writer by rapidly and effectively composing an essay on any topic (although perhaps not accurately). Sceptics argue that these tools are only good at organising words in a way that makes statistical sense, but cannot comprehend the meaning or determine the veracity of the claims they make.

It’s worth noting that ChatGPT’s data is limited until 2021, which means that the chatbot isn’t aware of news or events that have occurred since then. Consequently, certain questions you pose, like “Who won the World Cup in 2022?” won’t elicit a correct response.

The AI chatbot’s ability to produce entire articles within seconds can make it easier for students to cheat or bypass the need to learn how to write correctly, leading some educational institutions to block access to it. Another concern is that the chatbot may disseminate misleading information because of its lack of connectivity.

The demand for a ChatGPT text detector is growing due to increasing concerns about students using ChatGPT to cheat. OpenAI, the organisation that created ChatGPT, made a free programme available to address this issue. However, OpenAI’s “classifier” tool can only reliably identify 26 percent of content marked as “possibly AI-authored” as having been produced by AI. Furthermore, it delivers false positives 9 percent of the time, mistaking human-written content for AI-written content.

Before ChatGPT was made available to the public, Microsoft invested in OpenAI. In 2019, Microsoft made its initial investment of $1 billion. In the following years, it made additional investments totaling $2 billion. In January, Microsoft expanded its collaboration with OpenAI by making a multi-year, multi-billion dollar investment. Although neither company has acknowledged the investment sum, sources claim that it would total $10 billion over several years. As part of the deal, OpenAI will exclusively use Microsoft’s Azure cloud computing service to power all its workload for research, products and API services. Microsoft has also used its partnership with OpenAI to enhance its Bing search engine and improve its browser. On February 7, Microsoft unveiled a new Bing search engine that uses a next-generation OpenAI large language model customised specifically for search.

The author is a researcher, writer and analyst in the field of cyber security. He holds an LLB and an M Phil in cyber crimes. He is currently pursuing a PhD in computer science.