[Paper Review] Decoding ChatGPT: A Taxonomy of Existing Research, Current Challenges, and Possible Future Directions

Shahab Saquib Sohail, Faiza Farhat+6 more|arXiv (Cornell University)|Jul 26, 2023

Artificial Intelligence in Healthcare and Education198 references13 citations

TL;DR

A comprehensive literature review of 109 Scopus-indexed publications on ChatGPT, delivering a taxonomy of research domains, applications, challenges, and future directions.

ABSTRACT

Chat Generative Pre-trained Transformer (ChatGPT) has gained significant interest and attention since its launch in November 2022. It has shown impressive performance in various domains, including passing exams and creative writing. However, challenges and concerns related to biases and trust persist. In this work, we present a comprehensive review of over 100 Scopus-indexed publications on ChatGPT, aiming to provide a taxonomy of ChatGPT research and explore its applications. We critically analyze the existing literature, identifying common approaches employed in the studies. Additionally, we investigate diverse application areas where ChatGPT has found utility, such as healthcare, marketing and financial services, software engineering, academic and scientific writing, research and education, environmental science, and natural language processing. Through examining these applications, we gain valuable insights into the potential of ChatGPT in addressing real-world challenges. We also discuss crucial issues related to ChatGPT, including biases and trustworthiness, emphasizing the need for further research and development in these areas. Furthermore, we identify potential future directions for ChatGPT research, proposing solutions to current challenges and speculating on expected advancements. By fully leveraging the capabilities of ChatGPT, we can unlock its potential across various domains, leading to advancements in conversational AI and transformative impacts in society.

Motivation & Objective

• Provide a structured taxonomy of ChatGPT research across domains.
• Critically analyze methodologies and common approaches in existing studies.
• Map diverse applications of ChatGPT in healthcare, marketing/finance, software engineering, education, and more.
• Identify biases, trust issues, and ethical considerations surrounding ChatGPT.
• Propose future research directions and potential improvements for trustworthy, domain-specific AI usage.

Proposed method

• Adopt Kitchenham 2004 systematic review methodology.
• Apply inclusion/exclusion criteria to filter Scopus-indexed articles mentioning chatgpt/chat-gpt.
• Include 109 peer-reviewed articles up to March 25, 2023, authored by 349 researchers from 53 nations.
• Classify literature into domains and applications to build a comprehensive taxonomy.
• Analyze publication trends, collaboration networks, and dominant application areas.
• Synthesize findings to discuss biases, trust, and future directions.

Figure 1: Publications on ChatGPT on the basis of subject area

Experimental results

Research questions

• RQ1 RQ1: What is the current state of ChatGPT research, including architecture, advancements, and prime contributions?
• RQ2 RQ2: How diverse is the landscape of publications related to ChatGPT, and what are the recent trends?
• RQ3 RQ3: What are the various applications of ChatGPT across domains such as healthcare, marketing/financial services, software engineering, education, environmental science, and NLP?
• RQ4 RQ4: How can multimodal data be leveraged to enhance ChatGPT, and what are the key technical challenges?
• RQ5 RQ5: What are the main challenges, ethical considerations, potential risks, and ongoing research efforts in deploying GPT models in chatbot systems, and how are these addressed to ensure fairness, transparency, and human-centered design?

Key findings

• 109 articles analyzed, involving 349 authors from 53 nations.
• Medicine is the most represented field (23%), followed by social sciences (20%), and computer science (11%).
• The United States leads with 33 publications; top collaboration is US with 24 countries, Switzerland with 20, Australia with 19, and the UK with 18 collaborations.
• Publications mainly fall into three categories: evaluations of ChatGPT (68), predictions using ChatGPT (39), and reviews (10).
• Many studies assess ChatGPT for scientific writing and education, with growing attention to prompts engineering and diversity-promoting techniques.
• Ethical concerns, biases, and trustworthiness are recurrent themes, underscoring the need for further research and development in responsible AI deployment.

Figure 2: Top contributing countries