By Prakash Kumar
The making of public policy for a country is a complex task, and the complexity further increases if the country is large, populous and diverse. If a country does not attain the desired economic growth, questions are then raised whether inadequate public policies were framed or adequate ones were poorly implemented. Unfortunately, policymaking is not a scientific problem, which can be experimented with in a laboratory to test the hypothesis and measure its efficacy. Policymaking, by its very nature, is cumbersome, political, and bureaucratic.
How do we define a good policy-making process? Political science literature describes a “good policy-making process” as one committed to producing a high-quality decision. A good policy-making process requires up-to-date subject-matter knowledge, relevant data and the ability to analyse the data. Analytical tools enable policymakers to see the patterns in data. In addition to this, Artificial intelligence (AI) brings the ability to forecast the outcome, develop evidence-based programs and analyse effectiveness based on data, as AI systems can learn iteratively from both data and human interactions and build new knowledge and models based on the data.
Let us see how AI can help in policymaking. Policymaking is not a standalone activity but a multi-stage process of identification, trade-offs, formulation, adoption, implementation, and evaluation. At each stage, AI can help policymakers generate more value and have a better impact.
The first stage in policy formulation is identification of the problem. Also, policies made for one sector often have significant impacts on others. For example, a transport policy like the expansion of road network affects the environment. Policy-making, therefore, nearly always means trade-offs.
Policy-making processes and structures thus require deep analysis of information on such inter-sectoral impacts to enable fully informed choices between alternatives. AI tools can accelerate this process as they rapidly synthesize large amounts of data, detect patterns providing deep insights, and forecast the projected outcomes and benefits of policy options.
The next stage is the adoption of the drafted policy. AI can play an important role when such proposals are discussed and debated in the Parliament or State Assemblies. Armed with insights generated using AI during the prior stage, lawmakers will be better equipped to make more informed decisions.
Next comes the implementation of the policy. A policy only good if implemented well. Automation of the implementation process and near real-time analysis of feedback from the field is important for the efficient implementation of a policy. Analysis of grievances, queries and sentiments on social media by AI Tools provide quick insight into what is not working. This can enable tweaks/changes to be made in the policy to make it more impactful.
The last stage is evaluation. The faster the evaluation, the better gets the policy. AI tools bring the capability to speed up the assessment of components that need to change by identifying where a policy could be falling short or could be subject to fraud. For example, GST officials use AI-based tools to detect fraudulent refund transactions. Based on the outcome of this tool, policy changes were made in return processing and refunds.
AI, with many good features, also has its dark side as it is algorithm-driven. If AI systems are trained data sets that are biased, then they will inevitably embed data’s underlying social inequalities in their models. This is called logarithmic bias, which may reinforce discrimination if not handled properly. Thus, we need to be highly sensitive to data bias issues while using AI to abide by the principles of accountability, transparency, and fairness.
Conclusively, is this the right time to start using AI in policymaking? The answer could be Yes for many reasons. The governments are already using AI for improving service delivery and operations, so the time is ripe to use AI to support policymaking. With large-scale automation done by central and state governments, a huge amount of data is available in multiple sectors, which can be used to make new policies and evaluate existing policies. However, skills and capabilities need to be reinforced in the government for the use of AI. Taking a leaf from the experience of western countries, close partnership with institutions of learning is recommended as they have technical manpower and understanding of this technology.