The increasing prevalence of artificial intelligence (AI) applications, such as search engines, AI chatbots, social media content recommenders and image generation, are already impacting various aspects of our lives. People who possess an understanding of how AI functions will be better equipped to interact with the world and make informed decisions about utilizing and developing AI applications. Additionally, they will be capable of enhancing their critical thinking abilities, and will gain a deeper appreciation of how AI can be utilized to develop novel solutions to problems they may face.

This talk will be two-fold. In the first part, we will cut through the hype and clarify the definition of AI. Our discussion will revolve around the conventional computer programming approach, followed by an overview of how the machine learning example changes it. We will examine how models are trained in contrast to how programs are coded and highlight the essential role of data in this process. We will also explore how these models attempt to replicate the behavior of intelligent entities, by ""comprehending"" natural language and ""perceiving"" visual content to some extent.

The second part of the talk will focus on the challenges that may arise from data, including how AI applications can make mistakes due to poor data quality. Our examples will come from the language technology industry: language understanding and generation applications. Moreover, we will discuss the ethical and fairness considerations in developing responsible AI models. The talk will conclude by touching on the digital divide between languages and how it poses a threat to the survival of many of the world's 7000 spoken languages.

All in all, this talk aims to demystify and make AI more accessible to a wide audience, by clarifying its concepts and highlighting the importance of curated data in creating models that have real-world applications.