91�鶹������Ʒ����

Human beings depend on information for decision-making. What if we had perfect information, meaning that we have all the right information that we need to make decisions at the right time? All the information that we need for complex decision-making is not always available at the time needed; nor can the brain process at once all the information required in complex decision-making. Some actors think that computer technology like Artificial lntelligence (Al) can give us perfect information. So far, however, this is not humanly or technologically feasible.

AI nevertheless drives many of our most popular uses of technology by collecting, storing, processing and analysing data, and making decisions or helping people to make decisions. AI is not one unique tool but a collective system of software, hardware, algorithms, networks, etc. Algorithms are among the main drivers of AI systems. In the simplest sense of the term, an algorithm can be considered as a set of rules or list of instructions to complete a task or solve a problem. In that sense a recipe could be viewed as an algorithm. In the context of digital technologies, algorithms tells computers what steps to follow to carry out predetermined tasks or how to process data, and to make decisions based on given data. Algorithms are written by computer programmers. When many algorithms are combined into one system, they can perform relatively complex tasks or problem-solving.

Al can be applied in many fields. - from health diagnosis through to communications systems. Virtual assistants exist online and are embedded in technological tools to give selected answers to many of our questions including how to find a location or how to say something in a number of languages.

Robots are becoming ubiquitous. In disaster situations, AI applications can help humanitarian agencies get emergency supplies to the people who need it most urgently. Scientists use the speed at which AI converts data to information to address complex problems and make discoveries such as genome sequencing in much less time than before. As AI is integrated into an increasing number of technology solutions, it is seen as a general-purpose technology, a powerful tool with major impact on all aspects of our lives. The big question is who controls its development and deployment, and for what purposes. So far, it is the most powerful countries and industries, who, logically, aim to protect and promote their own interests and perspectives, which are not necessarily those of others. AI is also never neutral – it is engineered for particular purposes by humans with particular demographics and employment relationships, and the algorithms and data sets are always skewed in one way or another. Most virtual assistants are given a particular gender, and the programmed answers they provide reflect certain world views and biases. The language translation possibilities reflect dominant, not endangered, languages, while robots are geared for private rather than public use.

AI systems will typically demonstrate at least some of the following behaviours associated with human intelligence: planning, learning, reasoning, problem solving, knowledge representation, perception, motion, and manipulation and, to a lesser extent, social intelligence and creativity.  AI systems are driven by algorithms or sets of instructions that can be designed by humans or machines. There are many different types of Al, and no single definition. This module describes different dimensions of Al including machine learning, big data analytics, pattern recognition and cognitive systems and the difference between 'narrow' Al and 'general Al'.

Yet, the fact that computers can be programmed to copy “intelligent behaviour” and make independent decisions is of much concern. It raises questions about control. This in turn raises issues about human agency and the protection of fundamental rights including rights to freedom of expression, association and work. People are concerned about whether they will further lose their freedom to choose the type of content they want to see; whether Al will further deepen filter bubbles and information silos; ultimately reducing diversity and plurality of voices and content.

The UNESCO resource, , draws attention to some of these issues. This resource provides policy guidelines to tackle the persistence and severity of the gender gap in digital skills, and also looks at the “ICT gender equality paradox", which is UNESCO's finding that countries with high levels of gender equality have the lowest rates of women doing advanced studies in computer science or similar topics The publication also highlights for example, how the choice of product developers to use voices of young women in AI voice assistants perpetuates harmful gender biases, and offers recommendations to counter and reverse the widening of gender divides through and in AI. While some such voice assistants are becoming less stereotypically gendered, they still only serve a limited number of spoken languages.

The case of ‘Cambridge Analytica’ illustrates how AI-driven content moderation and curation can impact democratic systems. The “Cambridge Analytica” scandal was a case in which big data was used to influence voters without their knowledge. In some contexts, AI is seen as a tool for mass surveillance. Social media and other digital communications companies make extensive use of AI. Social media provides a use case study of how advanced machine learning impacts on user-generated content creation as well as marketing and purchasing decisions with both negative and positive implications. In education, the use of data analytics to profile learners is also perceived as having both positive and negative implications.

In parallel, the general lack of transparency in the design of algorithms and the data they access continues to cause concern.

The Ethical by Design ‘movement’ seeks to improve the design of AI in such a way as to combat algorithmic bias. ln the past, this has resulted in profiling and stereotyping of people, for example on the basis of factors such as race/ ethnicity, gender or language. “My Data belongs to me” is another coalition that seeks to help consumers and rights-holders take back control and ownership of personal data.

The opportunities that AI provides coupled with complex ethical and social concerns, highlight the need to balance innovation in AI with a human-centric approach anchored in clear ethical standards and societal goals. Further, they raise the question as to what type of knowledge, skills and attitude people need to purposefully and critically engage with AI systems. These competencies include both technical digital skills like programming/coding as well as soft competencies directly related to MIL such as critical thinking and civic engagement.

This module helps the reader to acquire a basic understanding about the technical operations and applications of AI systems as well as the economic and social context. It suggests how MIL competencies can enable more optimal use of AI in societies.