Kai's new watch is a bit intrusive, isn't it? Who gave it so much information? And above all, why does it want to organise Kai's life? He was so happy just doing his own thing!

The truth is that it's not very different from some of the devices we have at home or some of the assistants you have on your mobile phone, right? You can call it GPT, Chatbot, Co-pilot, Collaborator, Virtual Assistant, Smart Assistant, Virtual Guide, Digital Mentor…

How do these devices work? Where do they get their information from? And most importantly, where can I buy my own WAT? We don't have an answer to that last question. Remember that this is a comic created by the University of Granada and the DaSCI research centre. We don't sell WATs!

Article written by Rocío Romero Zaliz

Artificial intelligence (AI) assistants, such as Siri, Alexa, and Google Assistant, are now part of our everyday lives. Many people use them to streamline routine tasks, such as playing music or setting an alarm. These assistants are capable of recognising human speech, processing commands in real time, and responding to the user. In some cases, they even use an artificial voice to create a more pleasant user experience.

These assistants have evolved rapidly over the last decade. From the first voice recognition systems with limited capabilities to today's conversational assistants capable of maintaining complex and contextually coherent dialogues. This evolution has been made possible by advances in natural language processing, deep learning and the availability of large volumes of data for training these AIs.

Among AI assistants, we can find a particular class called wearable devices, such as smart watches or activity trackers. These devices integrate AI assistants to offer advanced functionalities that, in most cases, go beyond simple step tracking. For example, health monitoring systems in wearables can record parameters related to an individual's health, such as heart rate, blood oxygen saturation, sleep quality and even stress levels!

Thanks to this data, it is possible to design AI systems that analyse information and provide personalised recommendations. To do this, advanced AI technologies such as natural language processing and machine learning are used. If these devices are used over a long period of time, large volumes of data can be processed to enable early detection of diseases, identifying patterns and anomalies that could indicate potential health problems, alerting users and facilitating early intervention.

The latest advances in these devices include increasingly accurate sensors capable of detecting cardiac arrhythmias, blood glucose levels without the need for finger pricks, and even changes in skin conductivity that may indicate states of anxiety. Some manufacturers are working on systems that could detect early signs of depression based on patterns of activity, sleep, and heart rate variability.

AI assistants have also found their place in smart homes, where they control lighting, air conditioning, security and entertainment systems. The integration of these assistants with the Internet of Things (IoT) allows the creation of domestic ecosystems that learn from the habits of their inhabitants and adapt to their needs.

For example, an assistant could detect that we usually get up at 7:00 in the morning, so 15 minutes before that time, it begins to gradually raise the temperature of the home, prepares the coffee maker and adjusts the lighting to simulate a natural sunrise. These systems can also optimise energy consumption, reducing costs and environmental impact.

However, despite their benefits, AI assistants have some limitations and problems:

• Data dependency: AI assistants require a large amount of high-quality data to function properly. Obtaining this data can raise privacy concerns, as the collection and storage of personal data can be vulnerable to unauthorised access.
• Biases: Data quality is crucial, as AI systems are only as good as the data they are trained on. If this data is biased, the results will be too. For example, if most of the available medical data is from Caucasian patients and there is only a small amount of data from African American patients, the AI could make biased, erroneous, unfair, or even discriminatory decisions. These biases can manifest themselves in multiple ways: in facial recognition with poorer performance on people with darker skin, in hiring systems that favour profiles similar to those already in the company, or in credit risk prediction tools that perpetuate historical socio-economic inequalities.
• Transparency in the decision-making process: AI systems often function as a ‘black box,’ where there is data input and output, but what happens inside is unknown. It is difficult to understand how decisions are made, which can be problematic in important decisions, especially in health matters. A healthcare specialist would not trust this system if they do not know what the decision-making is based on, which generates mistrust and hinders its implementation.
• Privacy: AI assistants, being integrated into mobile devices, smart speakers or other devices, can have access to sensitive information such as personal data, GPS locations, Internet browsing habits, etc.
• Mass surveillance: Data collection by AI assistants can raise concerns about mass surveillance of individuals. On the one hand, it can help prevent crime and improve public safety, but on the other hand, a balance must be found that allows the benefits of technology to be exploited without sacrificing user privacy.

The future of AI assistants points towards greater personalisation and predictive capabilities. These systems are expected to be able to anticipate our needs based on behaviour patterns, contextual factors and personal preferences. Integration with other technologies such as augmented reality could create immersive experiences where our assistant guides us through physical environments with relevant information superimposed for us. 

For these technologies to reach their full potential, the ethical and technical challenges they pose will need to be effectively addressed. The creation of appropriate regulatory frameworks, the development of shared ethical standards, and the education of society on the responsible use of AI will be essential to ensure that AI assistants truly serve to improve our quality of life without compromising our fundamental rights.