Essay 5 ( AI watch : Defining Artificial Intelligence)

J R C     T E C H N I C A L      R E P O R T S

AI Watch

 Defining Artificial Intelligence

Towards an operational

definition and taxonomy 

of artificial intelligence

 Foreword 

This report is published in the context of AI Watch, the European Commission knowledge service to monitor the development, uptake and impact of Artificial Intelligence (AI) for Europe, launched in December 2018. AI has become an area of strategic importance with potential to be a key driver of economic development. AI also has a wide range of potential social implications. As part of its Digital Single Market Strategy, the European Commission put forward in April 2018 a European strategy on AI in its Communication "Artificial Intelligence for Europe" COM(2018)237. The aims of the European AI strategy announced in the communication are: 

● To boost the EU's technological and industrial capacity and AI uptake across the economy, both by the private and public sectors

 ● To prepare for socio-economic changes brought about by AI 

● To ensure an appropriate ethical and legal framework. 

Subsequently, in December 2018, the European Commission and the Member States published a “Coordinated Plan on Artificial Intelligence”, COM(2018)795, on the development of AI in the EU. The Coordinated Plan mentions the role of AI Watch to monitor its implementation. 

AI Watch monitors European Union’s industrial, technological and research capacity in AI; AI-related policy initiatives in the Member States; uptake and technical developments of AI; and AI impact. AI Watch has a European focus within the global landscape. In the context of AI Watch, the Commission works in coordination with Member States. AI Watch results and analyses are published on the AI Watch Portal (https://ec.europa.eu/knowledge4policy/ai-watch_en).

 From AI Watch in-depth analyses we will be able to understand better European Union’s areas of strength and areas where investment is needed. AI Watch will provide an independent assessment of the impacts and benefits of AI on growth, jobs, education, and society. 

AI Watch is developed by the Joint Research Centre (JRC) of the European Commission in collaboration with the Directorate‑General for Communications Networks, Content and Technology (DG CONNECT). This report addresses the following objectives of AI Watch: Developing an overview and analysis of the European AI ecosystem.

2.1 AI definitions 

Despite the increased interest in AI by the academia, industry and public institutions, there is no standard definition of what AI actually involves. AI has been described by certain approaches in relation to human intelligence, or intelligence in general. Many definitions refer to machines that behave like humans or are capable of actions that require intelligence (US NDAA, 2019; Russel and Norvig, 1955; McCarthy, 2007; Nilsson, 1998; Fogel, 1995; Albus, 1991; Luger and Stubblefield, 1993; Winston, 1992; McCarthy, 1988; Gardner, 1987; 1983; Newell and Simon, 1976; Bellman, 1978; Minsky, 1969; McCarthy et al., 1955). Since human intelligence is also difficult to define and measure, and although there have been different attempts of quantification (Gardner, 1983; 1987; Neisser et al., 1996), the objective definition of something as subjective and abstract as intelligence (Kaplan, 2016) falsely gives the impression of a precision that cannot be obtained. As a consequence, most definitions found in research, policy or market reports are vague and propose an ideal target rather than a measurable research concept. The oversimplification of the concept of intelligence that is needed in order to define, or even develop, AI is illustrated by Russel and Norvig (1985; 2010) and emphasised by the High Level Expert Group on Artificial Intelligence (HLEG, 2019) when focusing on rational AI and hence considering benchmark against an ideal performance. "A system is rational if it does the “right thing”, given what it knows" (Russel and Norvig, 1985; 2010).

Conclusion

The absence of a formal commonly agreed AI definition demanded the development of a process to establish a reference AI definition, and its subsequent operationalisation into a taxonomy and representative keywords, which can be adopted in the AI Watch framework and used in mapping and monitoring activities. The proposed iterative process includes three perspectives: policy and institutional, research, and market, in order to acquire a comprehensive overview about the AI domain. The AI definition adopted by the High Level Expert Group on AI is used as a baseline definition. It is selected based on the review of 55 relevant documents covering AI policy and institutional reports (including standardisation efforts, national strategies, and international organisations reports), research publications and market reports. An exhaustive list of the collected documents can be found in the report. The proposed operational definition is composed by a concise taxonomy characterising the core domains of the AI research field and transversal topics; and a list of keywords representative of such taxonomy. As AI is a dynamic field, we propose an iterative method that can be updated over time to capture the rapid AI evolution. While the baseline definition will be used as the general AI Watch definition of AI, the operational definition has a more functional use. Both the taxonomy and the list of keywords are essential to identify, map and characterise the worldwide AI landscape, one of the monitoring goals of AI Watch. The keywords are used in the initial phase to capture the relevant AI activities and the economic agents behind them. The main utility of the taxonomy is to classify AI activities, and will assist in the mapping of the AI landscape and the classification of economic agents’ areas of specialisation. Different uses of the keyword list are possible. A narrow use of the list, i.e. selecting only intrinsic-AI terms, allows to identify relevant AI activities, with an expected low proportion of false positives. When the objective is the categorisation of AI-related activities, a more comprehensive list is more suitable, in order to classify activities in their corresponding taxonomy domains. Valuable contributions of this work are: the collection of definitions developed between 1955 and 2019; the summarisation of the main features of the concept of artificial intelligence as reflected in the relevant literature; and the development of a replicable process that can provide a dynamic definition and taxonomy of the AI.

Full link text pdf :  https://eprints.ugd.edu.mk/28047/1/3.%20jrc118163_ai_watch._defining_artificial_intelligence_1.pdf