Siddharth Pai: Don’t be naive, Agentic AI won’t eliminate agency costs
Summary
Adopters of AI agents should be wary of risky ways in which these bots could veer off their objectives. The principal-agency problem of misaligned human motives is set to emerge in an artificial avataar that could prove even more costly.
In 1976, economists Michael Jensen and William Meckling—later my professors—introduced a theory that would fundamentally reshape corporate governance. Their insight was elegant and unsettling: whenever a ‘principal’ hires an ‘agent’ to act on its behalf, the agent’s behaviour may diverge from the principal’s interests.
This misalignment, whether stemming from perverse incentives, bad information or mere opportunism, gives rise to ‘agency costs.’ These costs extend beyond direct losses, encompassing expenditures on supervision, control and contract design—all intended to narrow the behavioural gap.
Jumbo July Offer is LIVE!
Get Upto 70% off on The Economist, WSJ & Mint Premium Plans
Subscribe nowAlready subscribed? Login
Premium benefits
200+ Premium Articles Weekly
WSJ, The Economist Access at < INR 340/month
Build Smart Finance Plans with Experts
Top Stock Recommendations Everyday
Expert-Backed Investment Growth
Access Mint Lounge, Newsletters & Insights
Catch all the Business News, Market News, Breaking News Events and Latest News Updates on Live Mint.
Download The Mint News App to get Daily Market Updates.
more
