What is ChatGPT?
ChatGPT (Chat Generative Pre-trained Transformer) is a chatbot developed by OpenAI in November, 2022.A chatbot is a software application for conducting an online chat through text or speech in place of a live human agent. OpenAI is a US artificial intelligence (AI) research laboratory committed to developing and promoting friendly AI to benefit mankind as a whole.
Thus, ChatGPT is basically an AI programme that has been designed for interactions in a conversational manner. It is a simulated chatbot primarily meant for customer service. The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit errors, contest incorrect premises, and reject inappropriate requests.
Following the launch of ChatGPT, the stakes of OpenAI rose substantially. It quickly caught the imagination of people for its detailed responses and articulate answers across diverse fields of knowledge.
Functions of ChatGPT
The basic function of a chatbot is to mimic a human conversation. ChatGPT has proved to be a versatile application in this format as it can write and debug computer programmes, compose music, teleplays and fairy tales, answer test questions to an amazing degree of accuracy, play games, simulate an ATM, etc.
ChatGPT is suitable for chatbots, AI system conversations and virtual assistants. Unlike the earlier version called InstructGPT, ChatGPT has also the inbuilt mechanism to minimise harmful and deceitful responses. It stores previous prompts from the same conversation, and has a filter mechanism to dismiss racist or sexist prompts.
Initial Performance
Following its launch in November, 2022 by the San Francisco-headquartered OpenAI, ChatGPT garnered over one million users within a month. Though free for users initially, the service will be monetized at a later stage. In fact some paid plans have already been introduced by OpenAI.
The initial reports suggest that the service works best in the English language. With respect to other languages, ChatGPT has varying degrees of success. It may be mentioned here that this chatbot has a language-based model that the developer fine-tunes for conducting conversations. The data source comprises textbooks, websites and articles which ChatGPT uses to model its own language for responding to human interaction.
However, the reviews have been generally positive so far. According to The New York Times, ChatGPT is “the best artificial intelligence chatbot ever released to the general public”. The Guardian also raved that it is able to generate “impressively detailed” and “human-like” text. The Atlantic included ChatGPT among the “Breakthroughs of the Year 2022” and said that it is part of “the generative-AI eruption” that “may change our mind about how we work, how we think, and what human creativity really is”.
‘Threat’ to Google
The New York Times reported that the “unexpected strength of ChatGPT” and the newly discovered potential of large language models to disrupt the search engine business have alarmed Google. Google CEO Sundar Pichai is believed to have “upended” and reassigned teams to aid in its AI products.
In fact ChatGPT has come as a veritable threat to popular AI-powered chatbots like Amazon Alexa, Google Bow and Siri. ChatGPT technology is superior because it is ‘trained’ to learn what people mean when they ask a question.
Negative Reviews
Economist Paul Krugman has gone on record stating that ChatGPT would hit the demand for knowledge workers. The Verge stated that the runaway success of ChatGPT proves that AI has gone mainstream.
Another economist, Tyler Cowen, has expressed concerns over the ChatGPT’s effects on democracy, citing the ability of the programme to write automated comments to affect the decision process of new regulations. The Guardian questioned whether any content found on the internet after ChatGPT’s launch “can be truly trusted” and called for government regulation.
Cyber security concerns
Check Point Research and other allied bodies pointed out that ChatGPT was capable of writing phising emails and malware. OpenAI (creator of ChatGPT) CEO Sam Altman himself wrote that advancing software could pose “(for example) a huge cyber security risk” and predicted that “we could get to real AGI (artificial general intelligence) in the next decade, so we have to take the risk of that extremely seriously”. Altman maintained that though ChatGPT is “obviously not close to AGI”, one should “trust the exponential”.
Training hiccups
A Time magazine probe revealed that OpenAI used outsourced Kenyan workers to label toxic content (sexual abuse, racism, etc) in order to build a safety system into the programme. The labels were used to train a model to detect toxic content in future. The outsourced workers were exposed to such toxic and dangerous content that they described the work experience as “torture”.
Limitations of ChatGPT
ChatGPT developer OpenAI has admitted that the programme “sometimes writes plausible-sounding but incorrect or nonsensical answers”. This behaviour technically called hallucination, is endemic to large language models. OpenAI further states that “training the model to be more cautious causes it to decline questions that it can answer correctly”. In addition, “supervised training misleads the model because the ideal answer depends on what the model knows, rather than what the human demonstrator knows”.
According to OpenAI, ChatGPT is “often excessively verbose and overuses certain phrases, such as restating that it is a language model...”. These issues are attributed to “biases in the training data (trainers prefer longer answers that look more comprehensive) and well-known over-optimization issues”.
Also, ChatGPT has limited knowledge of events post 2021.
Conclusion
ChatGPT, which is likely to revolutionise the tech world, can be tested for free on desktop and mobile websites. The application can be used in the browser from OpenAI’s ChatGPT website. The service may be availed for individual needs, expanded research, and other multiple activities.
Users have been impressed by ChatGPT’s ability to provide responses approximating to actual human conversations. This has also given rise to the possibility that this chatbot may eventually have the power to disrupt how humans interact with computers and change how information is retrieved.