The term artificial intelligence (AI) was officially coined by John McCarthy in 1956. AI is now considered the most important thing mankind is currently working on. Sundar Pichai, CEO of Google, believes that this technology is even more significant than fire and electricity.
So what is AI? AI is about building machines that can perform activities that currently can only be performed by brains – human or animal. At its very heart, it is the ability for machines to interpret and learn from data to make predictions. And they do so without being explicitly programmed for that.
There are various terms associated with AI, and within the AI community there is huge disagreement on what AI actually means and what its goals are. The past decades have witnessed dramatic scientific breakthroughs, especially in machine learning. While AI encompasses machine learning, machine learning is just one of the features of a fully intelligent general-purpose AI system. Machine learning allows expert systems to unlock abilities and faculties such as reasoning, perception, problem-solving, planning and more. These expert systems outperform human capacities, but only on specific tasks.
It is reasonable to expect that AI and machine learning will have a profound impact on our everyday lives and the skill sets companies look for in their employees. It will also affect learning and teaching. It is fair to ask how education will anticipate those changes. How successful education will be in doing so will depend on how well it manages to identify the frameworks within which the interaction between man and machine takes place.
Human activities & Artifical Intelligence
According to Leontiev’s activity theory, human processes can be examined from the perspective of three levels of analysis, which are linked to each other in a hierarchical way: activity, actions and operations.
The top of the hierarchy represents the activity and its underlying motive: it dictates the meaning of an activity. Any activity is performed through goal-oriented actions. The intermediate level represents actions, which are essentially ways of solving those problems that need to be solved in order to accomplish the activity. Operations, in turn, implement the actions using tools that are available; it is the bottom level. In other words, the level of activity provides the foundation for ethics. The actions implement the cognitive aspect of the activity and finally, operations are the repetitive routine skills to carry out the actions.
This three-level model of human processes provides a useful entry point for understanding AI and its potential impact on human activities. Societies and economies are reinventing themselves as users of new technologies with AI at the forefront. The impact of AI and automation on human activities is twofold: it can either transform them or replace them. Indeed, AI can support or substitute human activity when it comes to operations and actions, the two lowest levels of human processes.
Assuming that human-machine collaboration can relieve individuals from repetitive and, to a certain extent, cognitive tasks implies that people can – and should – shift their focus on the purpose and implications of their activity. This involves understanding the reason and meaning of their behavior and taking full responsibility for their actions. While today people are trained, through informal and formal education, to perform actions, tomorrow they will be required to take a step further. According to Gerbert (2018), “we need to prepare our people to ‘go meta’, leaving the constraints of the environment and moving to a higher abstraction or complementary level”, in other words, moving to their level of activity. The importance of educating people to reflect on the activity (in Leontiev’s sense) is exemplified by the consequences of the so-called algorithmic bias: by learning from data in order to mimic human decision processes, a machine will also replicate human biases. AI-based algorithms have thus been accused of fueling discrimination by widening inequality in education, reinforcing gender bias in recruitment and leading to racially biased verdicts in criminal cases, for instance. In other words, “just because a technology is accurate doesn’t make it fair or ethical” (Heilweil, 2020). 
This paradigm shift calls for an adjustment of workers’ skill sets. However, determining what competences need to be developed, and how, should stem from a holistic view of the systemic force of AI. AI should not be seen as a mere add-on tool, but must be understood as an intelligent system that can reshape human activities. Mastering its potential, therefore, goes hand in hand with the development of competencies that go well beyond the computer screen. Although education tends to focus mostly on the acquisition of technical competences, nine out of ten skills required in the future will be non-technical competencies, according to the World Economic Forum: 
- Technical competencies range from expert knowledge such as developer skills to general knowledge in algorithmics, AI’s limits and potential, and digitalization.
- Non-technical competencies are soft skills such as higher-order critical, creative, and innovative thinking, collective intelligence as well as emotional intelligence. As Hess (2017) puts it, “the new smart will be about high critical thinking and team collaboration”. 
Soft skills are crucial for one major reason: human-machine collaboration disrupts activity systems and therefore needs to be rethought and redesigned. Reflection should not be limited to AI’s uses for current processes, but should extend to imagining possible future (or even futuristic) scenarios. The role of education as an incubator of the competences needed to fuel this process is crucial.
AI & education
As part of the 2020 edition of Le Forum des 100 conference, a study  has been conducted to assess the extent to which upper-secondary and post-secondary schools in Suisse romande (French-speaking Switzerland) are preparing students – and therefore future workers – for the challenges and opportunities posed by AI. The findings of this study are insightful.
Firstly, the term “AI” is often misinterpreted. Teaching how to use a computer is not teaching about AI, nor is digitalizing education. In fact, AI principles and functions can be taught without the use of IT.
Secondly, there is some awareness that AI can serve schools as much as it can be taught at school. AI is reshaping the core foundations of education, teaching and learning. As was pointed out earlier, AI impacts human activity and this applies to teaching methods as well. Machine learning in education may indeed be an opportunity to walk away from a “one-size-fits-all” approach and to develop teaching tools that personalize learning and are tailored to individual needs and capabilities. In his article “How AI could personalize education”, Rouhiainen (2019) envisions “AI-based learning systems that would be able to give professors useful information about their students’ learning styles, abilities, and progress, and provide suggestions for how to customize their teaching methods to students’ individual needs”.  This would allow students experiencing learning difficulties, for instance, to acquire extra tutoring that specifically addresses the identified gap.
Last but not least, as human–machine collaboration enters the world of education, curricula will need to be rethought, and teachers and education designers will need to “go meta”. In its report on AI in education “Lead the leap”, Unesco (2019)  claims that education should not be about making the curriculum more technological, but about teaching more human-centric skills, i.e., shifting the focus to the top level of Leontiev’s pyramid, the activity. Unesco’s Beijing consensus on AI and education reminds us of the human-rights’ implications of “preparing all people with the appropriate values and skills needed for effective human–machine collaboration in life, learning and work, and for sustainable development”.  As the study on schools in Suisse romande shows, the importance of fostering this type of values and skills is generally overlooked.
The importance of soft skills
As contradictory as it may sound in the AI era, being smart will mean mastering a higher level of soft skills. On the one hand, understanding AI is key. On the other hand, developing soft skills is necessary to get the best out of this technology.
Although digitalization brings us one step closer to AI, too little emphasis is put today on the range of soft skills essential to make effective and responsible use of AI. Reflecting on the common perception that soft skills can be acquired mainly through life experiences, many universities and colleges see them as something that is “learned” only during internships. Explicitly developing soft skills in class would thus be a rather new responsibility for our education systems and an exciting opportunity to elevate the focus of educational programs from providing know-how to providing know-how-to-be.
 Heilweil, R. (2020, Feb. 18). Why algorithms can be racist and sexist. Vox.
 Hess, E. (2017, June 19). In the AI age, “Being Smart” Will Mean Something Completely Different. Harvard Business Review.
 Borrelly, C. and Sfreddo, C. (2020). La formation en réponse aux nouveaux défis économiques posés par l’intelligence artificielle. École hôtelière de Lausanne. Available from the authors.
 Rouhiainen, L. (2019, October 2019). How AI and Data Could Personalize Higher Education. Harvard Business Review.
 Unesco. (2019). International Conference on Artificial Intelligence and Education, Planning Education in the AI Era: Lead the leap – Final report.