ChatGPT: The Future Conversational Ai Chatbots

By Halima Ummi Ismail

Overview

Due to the advancement in technology, Artificial Intelligence (AI)  chatbots are now a part of human life. Although these chatbots have assisted humans in carrying out their daily activities, they are not without flows as majority of the responses are not totally accurate and the users are left with the tasks of sifting out the relevant information and discarding the rest. AI chatbots are now used in carrying out specific tasks such as writing copies for marketers, searching for books, and so on. The existence of chatbots is not a new phenomenon, however, ChatGPT is very distinct in terms of accuracy and analogy. This technology is very much faster and smarter than the humans.

Some weeks ago, various screenshots were posted online which were generated from ChatGPT, the latest AI language model. From the date it was launched, people have been testing it proficiency and it seems that this tool does not disappoint as it seems to have accurate answer to anything asked. It gives the right answers where necessary and refuses to do so where necessary based on the model it is developed on.

In November 2022, OpenAI which is a renowned San Francisco AI research and deployment company launched a language processing model known as “ChatGPT”. The chatbot which is said to be built on the company’s GPT (Generative Pre-trained Transformer ) 3.5 and a relative of the company’s InstructGPT has garnered the attention of over a million people in few days as they signed up to see what it is capable of. It interacts with people in a conversational manner while answering varieties of questions, admitting to mistakes, rejecting request that it considers inappropriate, and challenges the premises where it considers it to be incorrect.[i] A lots of screenshots of the conversations had with the chatbot were circulating the world via Twitter, and majority of people testified to the amazing nature of it responses which are very detailed.[ii] People are already referring to ChatGPT as the best AI chatbot that is available to the public at large.

The release of this chatbot by OpenAI has indeed cause a lot of excitement as numerous people across the globe including researchers awaits the outcome of the technology. People now look up to it in performing varieties of tasks while simplifying the workload that is done by humans across various fields. Obviously, this technology is here to stay and there is no going back.

ChatGPT has been used by many as it is deemed a clever chatbot whose ability supersede many of it kind. On a technical point, this chatbot is not just an ordinary one, it is a big deal. In order to fully expand its horizon, OpenAI has given users the opportunity of sending in their feedbacks to improve this technology. As such, the capacity of this AI keeps increasing every year and may go beyond human thinking within the next two years.

How Does ChatGPT works?

ChatGPT uses the features of a large language model. This model works based on a system of prediction generated from the large amount of data it was trained with to predict the appropriate words to use in constructing sentences.[iii] Interestingly, the more the data in it increases, the better the language model becomes. Large language models have limitations in understanding what human beings wants specifically. However, ChatGPT which seems to improve on these limitations  was trained by using feedback technique which is known as the Reinforcement Learning with Human Feedback (RLHF) training.[iv] This assist it in figuring out the expectations of humans and giving back to them the required results. It is said therefore, that this language model got a training that is revolutionary in nature.

In order to assist it in attaining the style of responding in a human form, ChatGPT got trained through massive data from sources such as Reddit as well as codes and information supplied from the internet. Indeed this chatbot is not like the ordinary chatbot used around as it detects the intentions of people while providing sincere and helpful feedbacks. So, it main focus is the preference of it users. However, there are certain questions that are not answered by it and in the construction of certain questions, it select the necessary part and discard the rest that makes no sense.

Uses Cases of ChatGPT

From the date it was launched ChatGPT has done incredibly well as professors, students, coders, engineers, writers and scientists have given it very difficult tasks to carry out which it did in few seconds. So the speed at which it provide those answers also matters a lot. This is a chatbot which is just on the outlook and is predicted to get smarter in the next one or two years, meaning it can totally outsmart humans.

ChatGPT has so far been used to write computer programs, compose songs, write and prepare plots for movies, answer assignment questions given to students, crack jokes, guess at diagnosis, suggest recipes, and even make comparison between different works of scholars. It is known to be very good in answering open ended type of questions and is a good tool for carrying out assignments. It has also been used to produce several essays in different languages across the world. The chatbot has been used to fix code with queries, coding of neural networks, and other computer programs. Surprisingly, the AI does most of the tasks given to it in few seconds which surpasses the human ability. In essence, the projects and works which will take humans so much time is being carried out by ChatGPT in few minutes.

ChatGPT is already taking it place in the marketing and content writing industry as marketers and copywriters are using it to generate very attractive and unique copies across various marketing niches. It thus, assist them in creating the structures and outline for the copy while enhancing their productivity. It is also a very good tool for giving researchers a background idea on a given topic. Unlike Google search which needs multiple structures to figure out, this chatbot goes straight to the point and produce the required answer in few seconds without delay. It has been tested and has given back very clear and attractive contents which surprised many in the industry. The adoption of this AI technology by writers, businesses, journalists and researchers will save them time by giving them the opportunity to focus their time and energy on other affairs.

ChatGPT and the future of EdTech

ChatGPT has drawn the attention of teachers and other stakeholders in the education sector from the day it was launched. Teachers in particular are afraid of the possibility of the AI disrupting students’ learning and aiding in exam malpractice and cheating. However, instead of relying on it disadvantages, recourse can be made to how the technology can be used in aiding learning and the proficiency of students. AI technologies are here to stay, and the earlier it is realized, the better for us all.

ChatGPT is one of those tools which is predicted to be very useful for education technology as it could be applied to aid learning in several ways.[v] It can serve as a personal tutor to students by giving them feedbacks or answers while providing students with explanations. It can also be used in reinforcing learning. Some other ways in which this AI can aid learning include providing the students with feedbacks on their assignments, it can be used to generate the appropriate lesson plan which the students can use personally to aid their learning, and so on.

ChatGPT and the Legal Profession

The legal profession has recently been disrupted by the evolution of AI as various AI tools are carrying out the work of lawyers accurately. The legal profession of the modern time is no longer the profession where basically everything was carried out through rigorous procedures. Instead of wasting time to search for cases using the old method, various AI tools are now used in searching for cases. Also, AI tools are used in drafting legal document and all other regulatory documents which were once the job of lawyers. As a matter of fact, the world is moving towards the possibility of having robots as lawyers and judges.

As for the use of ChatGPT in the legal profession, nothing is clear yet as the AI is not specifically created for that purpose. Personally, I have tried using the chatbot to analyse some cases one of which is the famous case of Salomon v Salomon and the answer given to me by it does not really look bad. It however, does not have the answer to some of my questions. Being a technology which is still evolving, this AI may become useful to lawyers as well.

Limitations

According to OpenAI, ChatGPT has certain limitations such as the possibility of providing the users with incorrect answers to their questions. The model is trained to be very cautious. It is also sensitive and may refuse to give answer to a particular query, but if the same question is changed with a slight twist of words or rephrasing, the chatbot can give the correct answer. The AI in some cases ask for clarifications whenever a query seems ambiguous to it.  Also, ChatGPT is said to have limited information as its knowledge is very much restricted to what it has learnt in the year 2021.

Another major limitations of OpenAI’s ChatGPT is the fact that the model has been made to reject inappropriate requests from users. Thus, it is programmed to reject giving answers to illegal activities and bias attitudes such as racisms and the likes. The purpose of doing that is to  prevent the model from being used for illegal and harmful purposes. However, users are trying hard to find a way of breaking the rule. Some people are also of the opinion that ChatGPT should not have such limitations.

Conclusion

In a nutshell OpenAI’s ChatGPT is a new AI model which is set to revolutionalise the world through its ability to analyse huge amount of texts in order to provide users with the required answer to their queries. The technology which was released few months ago has since then been tested by humans around the world in order to discover it true abilities. It has been used to carry out tasks which will ordinarily take humans so much time and efforts, and this technology does those tasks in few seconds. This technology is still in its development stage as there are high chances that it will indeed do more than what humans have seen it do, meaning that it will get smarter and better with time. And as for what it has in store for humanity, time will tell in due course. However, this AI model is definitely set to do wonders and will do nothing less than improving and reaching its full potentials.

Halima Ummi Ismail is a penultimate law student of Bayero University, Kano. She is an avid researcher with a keen interest in tech law. She can be contacted via email: ismailhalima99@gmail.com: LinkedIn: www.linkedin.com/in/halima-ummi-i-459267118

Author: Halima Ummi Ismail, School: Bayero University Kano, Phone Contact: 09078282927, Email Address: Ismailhalima99@Gmail.Com

[i] OpenAI ‘ChatGPT: Optimizing Language Models for Dialogue’ (2022) https://openai.com/blog/chatgpt/ accessed January 6, 2023

[ii] Kevin Roose ‘The Brilliance and Weirdness of ChatGPT’ (2022) The New York Times https://www.nytimes.com/2022/12/05/technology/chatgpt-ai-twitter.html   accessed January 6, 2022

[iii] Rebecca Heilwei ‘AI is finally Good at Stuff and that’s a Problem’ (2022) Vox https://www.vox.com/recode/2022/12/7/23498694/ai-artificial-intelligence-chat-gpt-openai  accessed January 6, 2022

[iv] Roger Montti ‘What is ChatGPT and How can You use it?’ (2022) Search Engine Journal https://www.searchenginejournal.com/what-is-chatgpt/473664/#:~:text=ChatGPT%20is%20a%20large%20language,in%20a%20series%20of%20words. Accessed January 6, 2022

[v] Colette Coleman ‘ChatGPT Gives Writing EdTech its Moment’ (2022) Forbes https://www.forbes.com/sites/colettecoleman/2022/12/19/chatgpt-gives-writing-edtech-its-moment/?sh=4be19a7652c1 accessed January 6, 2022