AI Ain’t Enough to Catch Defaulters

The Economic Times     26th August 2020     Save    

Context: The recent focus on Artificial Intelligence (AI) as the ultimate option the catch the defaulters in the banking system neglects the limits of AI and capabilities of Human Intelligence.

Advantages of AI:

  • Can Ascertain Ability to Pay: 
    • Smart algorithms, reinforced by machine learning (ML), may indeed add a speed and robustness to the loan approval process that’s more reliable than human agency.
    • AI may even enhance the discovery of creditworthy customers, especially micro, small and medium enterprises (MSMEs) who are denied by the banks today for lack of sufficient data.
  • Help modify future assumptions: based on macroeconomic factors like the state of the economy, and such specific indicators as a company’s finances and its immediate payment priorities.

Challenges with over-dependence on AI 

  • Difficulties of Predicting Behaviour: While predictions were made with the use of Big Data and larger populations, it often fails at the individual level.
    • It can only ascertain the ability to pay and not the intention to pay: 
      • Human intelligence may still be necessary to detect human delinquency.
      • It is doubtful that the woes of Infrastructure Leasing & Financial Services (IL&FS), Yes Bank or mounting Non-performing Assets (NPAs) have been forestalled by AI.

Way Forward:

  • ensure the rule of law: which cuts both ways:
    • Corporate houses should respect and conform to regulations that hold them accountable for their outstanding commitments.
    • Negligent bankers and regulators need to be held equally culpable in the event of an avoidable and catastrophic default.
      • For E.g. The Reserve Bank of India can’t be allowed to surrender its responsibility to a program
  • Regulatory Investments: In addition to technological investments, the health of the financial system depends on resource-intensive investments in regulatory oversight and enforcement.