Economists Discuss AI Alignment Problem in Monetary Policy

• A recent report highlighted the “alignment problem” associated with artificial intelligence.
• Economist James Broughel has presented this problem in the context of monetary economics.
• Broughel suggested a committee to oversee the functioning of an AI system.

Artificial Intelligence (AI) researchers and economists have highlighted the risks posed by the use of AI in our daily lives. According to the experts, policymakers need to address the “alignment problem,” which arises when we depend on artificial intelligence without aligning it with our values and goals.

According to a report by Forbes, the alignment problem is arguably the largest challenge in the artificial intelligence space. AI algorithms operate on a set of rules which direct them to their goal. This may sometimes lead to an outcome that was not initially desired, but the input to the AI algorithm made it appear so.

Economist James Broughel presented the AI problem in the context of monetary policies. He cited the example of a rules-based regime where the central bank follows a defined set of rules. Such a regime helps with consistency and predictability in the market but isn’t very adaptive to changing circumstances.

On the other hand, a discretionary regime has a central committee that exercises a considerable level of flexibility to adjust its monetary policy based on the economic landscape and changing circumstances. In this case, the central bank is usually able to address unexpected activity and enforce an appropriate policy based on the situation.

Broughel believes that a goal-oriented artificial intelligence algorithm may lead to unintended consequences in the process of achieving its goal. He argued that such AI facilities should have a killswitch-like mechanism to shut them down in case of an emergency. Twitter CEO Elon Musk has shared earlier his concerns and anxieties about AI’s “great danger,” pointing out that “AI stresses me [him] out.”

Another solution would be to appoint a manager or committee to oversee the operations of AI systems. This group would be responsible for suspending the AI system if it strays from its objective. The economist added that an emergency switch would have to be programmed into the AI algorithm to make sure that it adheres to said committee.

Post Views: 19