A large language model (LLM) chatbot, ChatGPT, was developed by an artificial intelligence company named OpenAI based in San Francisco. As the name suggests, ChatGPT is a form of chatbot which has the capability of interacting in a surprisingly conversational and dialogue form. It provides responses which can appear to have been provided by a human.
Since ChatGPT is a large language model, therefore, it is first essential to know what does a large language model means. These models are specifically created with enormous amount of data in a way so as to accurately predict which word comes next in a sentence.
These large language models predict the upcoming word in a series of words in a sentence and the following sentences. It may be called an auto-complete which writes down paragraphs and pages of content by the predicting the words to follow. But the problem with this lies in the fact that the large language models are mostly limited, which is why they are incapable of understanding and analyzing exactly what a human mind wishes to write.
ChatGPT comes into play here, as it has been built with the improved state of the art, with the Reinforcement Learning with Human Feedback (RLHF) training. ChatGPT has been trained to attain a way of responding in a human way and has also been designed to learn a dialogue, which is why it has the ability to converse and respond like the humans. Basically, it was trained using a technique called Reinforcement Learning with Human Feedback (RLHF) that makes the use of human feedback that helps ChatGPT to comprehend the task of following instructions given by a person and in turn, provide acceptable responses.
Advantages and uses of ChatGPT
Personalized customer service is one of the main benefits of utilizing ChatGPT. Without human assistance, ChatGPT can offer 24/7 rapid and effective customer service which may utilize consumer information to personalize conversations and enhance users’ overall experiences, which will also increase customer satisfaction and cut down on wait times. It goes without saying that by doing routine jobs like responding to commonly asked queries, these AI chatbots enhance productivity by freeing up human employees to concentrate on more difficult tasks. Additionally, organisations may save a lot of money by taking the help of a chatbot like ChatGPT. It reduces corporate expenditures by doing away with the requirement for human customer support personnel. Talking about benefits of ChatGPT, it also helps in the medical industry by summarizing huge volumes of medical records which are based on the patients’ family history, lab results, symptoms, and in turn benefitting the industry by providing a more effective diagnosis.
Disadvantages of ChatGPT
One of the major disadvantages of ChatGPT is its possible use in phishing and malware attacks. Phishing emails and messages may be written convincingly by hackers using AI-powered chatbots, such as ChatGPT, making them harder to spot and avoid. These chatbots may mimic writing styles and make it seem as though they are coming from a reliable source, duping users into disclosing private information or installing malicious software. Additionally, the ability of chatbots to create malware programmes and execute cyber-attacks may help malware spread quicker to a large number of users, ultimately making it more challenging to spot and halt the attack.
Apart from these malware and phishing concerns, another concern which raises the alarm is privacy issues. The usage of ChatGPT in customer service may raise issues regarding data privacy and the collection, storage, and use of personal sensitive information. These chatbots could gather and retain information like name, email address, and other information that could be exposed to data breaches and other cyber-attacks.
Legality of ChatGPT
Speaking about legality, as of now, there is no specific legislation pertaining to it. ChatGPT is banned at numerous schools and institutions, notably in India, France, and the US. This ban comes from the concern that students are gaining an unfair advantage by getting a detailed response on the topics for articles and assignments. This is resulting in their knowledge being underdeveloped by misusing ChatGPT. However, if we look further, a Columbian judge had recently used ChatGPT to dictate a judgment. Hence, legal implications of ChatGPT are still in debate and not yet decided. For the inclusion of ChatGPT in the Indian legal framework, the legal IT framework needs to be amended accordingly. Regarding the issue of liability for content created by ChatGPT, given that the AI tool is typically used in the manner of a search engine and receives input from users, ChatGPT may be considered an “intermediary” as defined by the Information Technology Act 2000. It will, therefore, be qualified for the safe harbour protection that is granted to intermediaries. The liability for the content used to train ChatGPT, there is of course, uncertainty on how such content has been secured, including the remuneration for such access. Also, in the case where ChatGPT is being used very often, concerns about the copyrighted materials, used to train it, are expected to come up.
It may be stated that ChatGPT is a wonderful technology, which is a prime illustration of how quickly AI can reach to the masses. However, there is still a plethora of legal issues and concerns that could prevent its widespread use. This AI generated chatbot also comes with its own limitations, for instance, it can only provide answers which are based on the data it has been trained on. Hence, to conclude, it would be rightful to state that the use of such AI has to be regulated by bringing amendments in the IT laws and to develop a highly trained unit, dedicated for the monitoring of such AI. As much helpful ChatGPT is in our day-to-day activities, its related concerns also needs addressing. Hence, placing complete reliance on AI may turn out harmful in terms of privacy breach or data theft.
Author – Shyamli Shukla
Co-Author – Rashi Srivastava