The Centaur model, named after the mythological hybrid of man and horse, represents a paradigm for collaboration between human and artificial intelligence. This concept, originally introduced by Garry Kasparov in the context of chess (Kasparov, 2017), has evolved into a broader model for the integration of human and machine capabilities in different domains [1].
Theoretical roots
The theoretical roots of the Centaur model can be traced back to Licklider’s (1960) concept of ‘human-computer symbiosis’ [2]. Licklider foresaw a future in which ‘human brains and computing machines will be very closely coupled’ and ‘the resulting partnership will think much more efficiently than a human being alone has ever thought.’ This vision forms the basis for the modern understanding of human-AI collaboration.
Brynjolfsson and McAfee (2014) expand on this concept in their work ‘The Second Machine Age’ [3]. They argue that the combination of human and artificial intelligence can lead to achievements that neither humans nor machines alone could achieve. This synergy is at the centre of the Centaur model.
While the term ‘cyborg’ (from ‘cybernetic organism’) refers to a biological being that has been enhanced by technological components (Clynes & Kline, 1960) [4], and ‘android’ describes an artificial, human-like robot (Ishiguro, 2006) [5], the concept of the ‘centaur’ stands for a symbiotic collaboration between man and machine without physical fusion.
With her ‘Cyborg Manifesto’, Haraway (1985) shaped the cultural-theoretical debate on human-machine hybrids [6], while the Centaur model according to Kasparov (2017) deliberately chooses the metaphor of the mythological hybrid being to emphasise the synergetic partnership while preserving the autonomy of both actors [1]. As predicted by Licklider, this form of cooperation makes it possible to utilise the complementary strengths of both intelligences without compromising the physical or cognitive integrity of humans (Rahwan et al., 2019) [7].
Effectiveness of the model
Empirical studies have investigated the effectiveness of the Centaur model in various areas. Jarrahi (2018) analysed the role of AI in organisational decision-making processes and found that the combination of human intuition and machine analysis capabilities can lead to improved decisions [4]. These results support the core thesis of the Centaur model that the synergy of humans and machines leads to superior outcomes.
In the field of medical diagnostics, studies have shown that AI-assisted doctors can make more accurate diagnoses than either AI systems or doctors alone (Liu et al., 2019) [5]. These findings emphasise the potential of the Centaur model in critical areas of application.
Despite promising results, the implementation of the Centaur model faces challenges. In their study, Grønsund and Aanestad (2020) identify difficulties in integrating AI into existing workflows and emphasise the need to develop new ‘human-in-the-loop’ configurations [6]. This suggests that the successful implementation of the Centaur model requires careful design of the human-machine interface.
Dellermann et al (2021) present a taxonomy for the design of hybrid intelligence systems, emphasising the complexity of the task of effectively combining human and artificial intelligence [7]. They argue that a deep understanding of both the human and machine components is required to realise the full potential of the Centaur model.
Ethical issues
The increasing integration of AI into human decision-making processes raises many ethical questions. In their study of ‘machine behaviour’, Rahwan et al. (2019) discuss the need to better understand the interactions between human and machine behaviour in order to minimise potential negative effects [8].
The Centaur model offers perspectives for the future of work and innovation. However, it requires further research to validate its applicability in different domains and to develop optimal implementation strategies. Future studies could focus on the development of specific training methods for human-AI teams, the refinement of human-machine interfaces and the investigation of long-term effects of Centaur configurations on human capabilities and organisational structures.
The Centaur model represents an approach to utilising the complementary strengths of human and artificial intelligence. While empirical findings demonstrate its effectiveness in various areas, challenges remain in its practical implementation. Further research is needed to realise the full potential of this model and to fully understand its ethical and societal implications.
References:
[1] Kasparov, G. (2017). Deep Thinking: Where Machine Intelligence Ends and Human Creativity Begins. PublicAffairs.
[2] Licklider, J. C. R. (1960). Man-computer symbiosis. IRE transactions on human factors in electronics, (1), 4-11.
[3] Brynjolfsson, E., & McAfee, A. (2014). The second machine age: Work, progress, and prosperity in a time of brilliant technologies. WW Norton & Company.
[4] Clynes, M. E., & Kline, N. S. (1960). Cyborgs and space. Astronautics, 14(9), 26-27, 74-76.
[6] Ishiguro, H. (2006). Android science: Conscious and subconscious recognition. Connection Science, 18(4), 319-332.
[6] Haraway, D. (1985). A Manifesto for Cyborgs: Science, Technology, and Socialist Feminism in the 1980s. Socialist Review, 15(2), 65-107.
[7] Rahwan, I., Cebrian, M., Obradovich, N., Bongard, J., Bonnefon, J. F., Breazeal, C., & Wellman, M. (2019). Machine behaviour. Nature, 568(7753), 477-486.
[8] Jarrahi, M. H. (2018). Artificial intelligence and the future of work: Human-AI symbiosis in organizational decision making. Business Horizons, 61(4), 577-586.
[9] Liu, X., Faes, L., Kale, A. U., Wagner, S. K., Fu, D. J., Bruynseels, A., … & Denniston, A. K. (2019). A comparison of deep learning performance against health-care professionals in detecting diseases from medical imaging: a systematic review and meta-analysis. The lancet digital health, 1(6), e271-e297.
[10] Grønsund, T., & Aanestad, M. (2020). Augmenting the algorithm: Emerging human-in-the-loop work configurations. The Journal of Strategic Information Systems, 29(2), 101614.
[11] Dellermann, D., Calma, A., Lipusch, N., Weber, T., Weigel, S., & Ebel, P. (2021). The future of human-AI collaboration: A taxonomy of design knowledge for hybrid intelligence systems (Version 1). arXiv.
[12] Rahwan, I., Cebrian, M., Obradovich, N., Bongard, J., Bonnefon, J. F., Breazeal, C., … & Wellman, M. (2019). Machine behaviour. Nature, 568(7753), 477-486.