Challenges to consider and assess
Research in ethical and regulatory considerations for AI – driven health interventions is lacking as supported by (Schwalbe and Wahl 2020).
The need for more research in the practical application of AI-driven interventions in the health sector is also echoed by (Sharma, Yadav, and Chopra 2020), with a highlight to consider research on digital governance.
According to (Gasser and Almeida 2017), the scale, heterogeneity, complexity, and degree of technological autonomy requires new thinking about policy, law and regulation.
When decisions are made by AI systems that affect human lives like in medicine, there is an emerging need for understanding how such systems work (Barredo Arrieta et al. 2020), hence the need for transparency.
To consider AI as a solution to itself by embracing some of the AI techniques proposed by (Milano, O’Sullivan, and Gavanelli, 2014) to aid in the policy-making process such as decision support and optimisation techniques, game theory, data and opinion mining and agent-based simulation.
The potential of Agent based simulation as a visualisation tool towards measuring the AI-driven intervention values and ethics considerations.
(Schwalbe and Wahl, 2020) proposed establishment of guidelines for development, testing, and use as well as development of a user-driven research agenda to establish equitable and ethical use of AI-driven health interventions, which are approached from a needs-based rather than from a tool-based angle. Such regulatory, ethical and economic standards as well as guidelines are to safeguard the interests of LMICS (Schwalbe and Wahl, 2020).
The PPI can potentially be adopted as a viable approach at a sectoral level of public health AI policy formulations as supported by(Guerrero, 2020)
AI Governance can assist in embracing the behaviour of each stakeholder through the representation of their respective norms and values in an agent-based model.
Many stakeholders’ entails competing viewpoints (Milano, O’Sullivan, and Gavanelli, 2014) the only solution to enable dialogue and reach a desirable conclusion, is when all viewpoints (data privacy, data security, data bias) can be assessed in a transparent, inclusive and decisive manner.
The interaction of various components through agent-based modelling could help minimise the black-box (Reisman et al., 2018) effect. Agent based modelling will allow representation of various key player’s attributes in a transparent and unbiased approach.
How can governments evaluate the impact of AI-driven intervention in the health sector as an application example for consideration.
1. Barredo Arrieta, Alejandro, Natalia Díaz-Rodríguez, Javier Del Ser, Adrien Bennetot, Siham Tabik, Alberto Barbado, Salvador Garcia, et al. 2020. ‘Explainable Artificial Intelligence (XAI): Concepts, Taxonomies, Opportunities and Challenges toward Responsible AI’. Information Fusion 58 (June): 82–115. https://doi.org/10.1016/j.inffus.2019.12.012.
2. Gasser, Urs, and Virgilio A.F. Almeida. 2017. ‘A Layered Model for AI Governance’. IEEE Internet Computing 21 (6): 58–62. https://doi.org/10.1109/MIC.2017.4180835.
3. Guerrero, Omar A. 2020. ‘Policy Priority Inference’. The Alan Turing Institute. Policy Priority Inference (blog). 2020. http://oguerr.com/ppi/.
4. Milano, Michela, Barry O’Sullivan, and Marco Gavanelli. 2014. ‘Sustainable Policy Making: A Strategic Challenge for Artificial Intelligence’. AI Magazine 35 (3): 22–35. https://doi.org/10.1609/aimag.v35i3.2534.
5. Reisman, Dillon, Jason Schultz, Kate Crawford, and Meredith Whittaker. 2018. ‘Algorithmic Impact Assessments: A Practical Framework for Public Agency Accountability’.
6. Schwalbe, Nina, and Brian Wahl. 2020. ‘Artificial Intelligence and the Future of Global Health’. The Lancet 395 (10236): 1579–86. https://doi.org/10.1016/S0140-6736(20)30226-9.
7. Sharma, Gagan Deep, Anshita Yadav, and Ritika Chopra. 2020. ‘Artificial Intelligence and Effective Governance: A Review, Critique and Research Agenda’. Sustainable Futures 2: 100004. https://doi.org/10.1016/j.sftr.2019.100004.