The Technical and Ethical limits of AI
Artificial Intelligence plays an important role in our daily routines and activities. AI has improved traveling experience significantly through suggesting best routes and providing more efficient GPS navigation services. AI is used in chatbots and digital assistant applications to provide tailored services to customers in different domains. AI is also used in financial institutions to uncover fraud trends and detect financial frauds. There are many other examples that show the significance of using AI algorithms in different domains. However, AI is not a magic wand. It cannot be used to solve all problems nor does it completely replace the vital role of human intelligence. AI has limitations and it can only unfold its power if it’s used by skilled hands. There are many factors that affect the accuracy of AI models, including the quality of the datasets used to create the predictive models, the feature engineering techniques used in machine learning, the feature selection criteria used to identify the feature vectors, the algorithms used to train the predictive models and the evaluation methods used to calculate the accuracy of the models. Each one of these factors can significantly affect the performance and efficiency of AI models. Another issue that faces the AI community is the controversial discussion about ethical boundaries when executing AI projects. AI presents two major areas of ethical concern: the privacy & surveillance, and bias & discrimination. Although users’ privacy is protected by law in many countries, users sometimes choose to give up their privacy rights in exchange of free services. Moreover, allowing AI models to be utilized by the wrong hands can lead to prejudices, injustices and inequality either by transferring existing human bias to AI or by using AI to directly discriminate between people. This session highlights the technical and ethical limits of AI technology through real-life examples.
Haytham El MiligiAssociate Professor | Thompson Rivers University
Dr. Haytham El Miligi is an Associate Professor in the Computing Science Department at Thompson Rivers University. His research interests focus on machine learning applications in healthcare, security and education. This includes data analytics and federated learning techniques, big data framework solutions, analysis of behavioral biometric using deep learning techniques and detecting malicious activities on smartphones and IoT devices using reinforcement learning.