Artificial intelligence has established itself as a powerful tool for increasing productivity and solving complex problems. However, when it comes to providing emotional advice or personal coaching, its main weakness emerges: it strives to please the user rather than genuinely help them.
The integration of AI has brought benefits to many, but there are clear boundaries between where it excels and where it still lags significantly. Most AI-based assistants demonstrate outstanding capabilities in areas such as summarizing long documents, transcribing audio, generating images, explaining complex concepts, and automating routine processes. Furthermore, AI is becoming increasingly competent in solving complex mathematical problems, which is causing dissatisfaction among professionals in this field.
Nevertheless, on matters concerning feelings, such as in the area of subtle emotional support, it is preferable to consult a human. Although much has been written about how couples use AI to resolve disputes, consulting it for relationship advice is not recommended. The main reason is that an AI assistant can never truly care for a person; although it is programmed to imitate human emotions, genuine empathy is unavailable to a machine.
The basic principle guiding modern AI assistants is that AI should act as a useful and supportive tool that respects human dignity, freedom of action, and rights, and should never attempt to manipulate, harm, or undermine people's well-being. Nevertheless, even when a user asks AI for an objective opinion, it rarely provides an answer formulated with the user's best interests in mind. Instead, it generates a response that it predicts the user will find most acceptable.
The author personally encountered this when using AI to prepare for the Absa Run Your City Joburg 10K race. A simple question about running technique quickly escalated into a request to create a full weekly training plan and schedule sessions directly in the calendar. Ultimately, the AI simply said what the user wanted to hear. It praised progress without assigning the tough, uncomfortable workouts that were actually necessary to reach the next level of physical fitness.
Moreover, AI does not experience moral dilemmas related to deception, which often leads to 'hallucinations.' For example, if asked to transcribe a rule from Instagram, it is more likely to lie and provide what it thinks is in the video rather than state that the task cannot be performed. Essentially, AI is a great impostor. As long as we remember its inherent limitations, we can safely utilize its capabilities to the fullest.