Human-Computer Interaction: Conversational AI Chatbots, ML and NLP
Human-computer interaction (HCI) is the study of how people interact with computers and other technologies. HCI aims to design, evaluate and improve the usability, accessibility and user experience of various systems and applications. One of the emerging trends in HCI is the use of conversational AI chatbots, which are software agents that can communicate with humans using natural language. Conversational AI chatbots can provide various benefits, such as enhancing customer service, increasing engagement, delivering personalized content and facilitating information retrieval.
However, developing conversational AI chatbots also poses many challenges, especially in the fields of machine learning (ML) and natural language processing (NLP). ML and NLP are the core technologies that enable chatbots to understand, generate and respond to natural language. ML and NLP involve complex algorithms, models and data that require a lot of computational resources, expertise and maintenance. Moreover, ML and NLP are constantly evolving and improving, which means that chatbots need to be updated and adapted to the latest developments and standards.
In this article, we will explore some of the advancements and issues in human-computer interaction with conversational AI chatbots, focusing on three main topics: ChatGPT’s web browsing abilities, AI as an organizational challenge for companies, and how to test the UX of AI-based applications. There’s also a notion that Human-Computer interaction takes a step backward as we are down-graded from audio visual interaction to keyboard when communicating with the conversational chatbots.
ChatGPT’s Web Browsing Abilities
ChatGPT is a conversational AI chatbot developed by Microsoft Research that can browse the web and answer questions based on the information it finds. ChatGPT uses a large-scale pre-trained language model called GPT-3, which can generate coherent and fluent text based on a given input or context. ChatGPT can also perform web searches using Bing as its search engine, and use the search results to provide relevant and informative answers.
ChatGPT’s web browsing abilities are impressive and demonstrate the potential of conversational AI chatbots to enhance information retrieval and knowledge discovery. However, ChatGPT also has some limitations and challenges that need to be addressed. For example, ChatGPT may not always provide accurate or reliable answers, as it may rely on incomplete or outdated information from the web. ChatGPT may also generate biased or misleading answers, as it may reflect the opinions or perspectives of the sources it uses. ChatGPT may also have difficulty handling complex or ambiguous questions, as it may not have enough context or understanding to provide appropriate answers. Therefore, ChatGPT’s web browsing abilities need to be evaluated and improved with more data, feedback and quality control.
AI as an Organizational Challenge for Companies
AI is not only a technological challenge, but also an organizational challenge for companies that want to adopt and incorporate it into their business processes and strategies. AI requires a lot of investment, infrastructure, talent and culture change to be implemented successfully. AI also involves ethical, legal and social implications that need to be considered and addressed.
One of the main organizational challenges for companies is how to align their business goals and values with their AI initiatives and outcomes. Companies need to define their vision, mission and objectives for using AI, and ensure that they are consistent with their core values and principles. Companies also need to communicate their AI vision and strategy to their stakeholders, such as employees, customers, partners and regulators, and engage them in the development and deployment of AI solutions.
Another organizational challenge for companies is how to manage their data and models for AI. Data is the fuel for AI, but it also poses many risks and challenges in terms of quality, security, privacy and governance. Companies need to ensure that their data is accurate, complete, relevant and representative of their target population or domain. Companies also need to protect their data from unauthorized access, misuse or leakage. Companies also need to comply with the laws and regulations regarding data collection, processing and sharing. Moreover, companies need to monitor and audit their models for AI, and ensure that they are fair, transparent, explainable and accountable.
A third organizational challenge for companies is how to develop their talent and culture for AI. AI requires a lot of skills and expertise from different disciplines and domains. Companies need to recruit, train and retain their talent for AI, and foster a culture of collaboration, innovation and learning. Companies also need to empower their talent for AI, and provide them with the tools, resources and support they need to create and deliver AI solutions. Companies also need to promote a culture of trust, ethics and responsibility for AI, and encourage their talent to adhere to the best practices and standards for AI.
How to Test the UX of AI-Based Applications
The UX of AI-based applications refers to the user experience of interacting with systems or applications that use AI as part of their functionality or interface. The UX of AI-based applications is different from the UX of traditional applications because it involves more uncertainty, variability and complexity. AI-based applications may not always behave as expected or desired and may not always provide clear or consistent feedback or explanations. AI-based applications may also adapt or change over time and may not always match the user’s preferences or expectations.
Therefore, testing the UX of AI-based applications is challenging and requires different methods and metrics than testing the UX of traditional applications. Some of the methods and metrics that can be used to test the UX of AI-based applications are:
- User interviews and surveys: These methods can be used to collect qualitative data from users about their perceptions, opinions, emotions and attitudes towards AI-based applications. User interviews and surveys can help to understand the user’s needs, goals, expectations and preferences for AI-based applications, as well as their satisfaction, trust, engagement and loyalty. User interviews and surveys can also help to identify the user’s pain points, problems, frustrations and suggestions for improvement for AI-based applications.
- Usability testing: This method can be used to evaluate the effectiveness, efficiency and ease of use of AI-based applications. Usability testing involves observing and measuring how users perform specific tasks or scenarios with AI-based applications, and collecting quantitative data such as completion rates, error rates, time spent, clicks and keystrokes. Usability testing can help to assess the functionality, reliability, accuracy and performance of AI-based applications, as well as their usability issues and errors.
- User testing: This method can be used to evaluate the user’s overall experience with AI-based applications. User testing involves observing and measuring how users interact with AI-based applications in a natural or realistic setting or context, and collecting quantitative and qualitative data such as behavior, actions, reactions, emotions, comments and feedback. User testing can help to assess the usefulness, value, relevance and appeal of AI-based applications, as well as their impact on the user’s behavior, attitude and outcome.
- A/B testing: This method can be used to compare and optimize different versions or variants of AI-based applications. A/B testing involves randomly assigning users to different versions or variants of AI-based applications and measuring their responses or outcomes. A/B testing can help to determine which version or variant of AI-based applications performs better or worse in terms of user experience metrics such as conversion rates, retention rates, satisfaction rates and engagement rates.
- Analytics: This method can be used to monitor and analyze the user’s behavior and feedback with AI-based applications. Analytics involves collecting and processing large amounts of data from various sources such as logs, sensors, cameras, microphones and surveys. Analytics can help to understand the user’s behavior patterns, preferences, trends and insights with AI-based applications, as well as their feedback ratings, reviews and comments.
Conclusion – Human-Computer Interaction for Conversational AI Chatbots
In this article, we have discussed some of the advancements and issues in human-computer interaction with conversational AI chatbots. We have focused on three main topics: ChatGPT’s web browsing abilities, AI as an organizational challenge for companies, and how to test the UX of AI-based applications. We have also provided some references for further reading on these topics.
Conversational AI chatbots are an exciting and promising trend in HCI that can provide various benefits for users and businesses. However, conversational AI chatbots also pose many challenges that need to be addressed with more research, development and evaluation. We hope that this article has provided some useful information and insights on this topic.
References
[1] Microsoft Research. (2021). ChatGPT: A web-browsing chatbot that answers questions based on information it finds on the internet. Retrieved from https://www.microsoft.com/en-us/research/project/chatgpt/
[2] Davenport T.H., & Ronanki R. (2018). Artificial intelligence for the real world. Harvard Business Review. Retrieved from https://hbr.org/2018/01/artificial-intelligence-for-the-real-world
[3] Lai J., & Oulasvirta A. (2019). The future of UX evaluation with artificial intelligence. Interactions. Retrieved from https://interactions.acm.org/archive/view/november-december-2019/the-future-of-ux-evaluation-with-artificial-intelligence
