2 min

Salesforce has unveiled its new AI Economist. The purpose of the AI is to help leaders create tax policies that achieve specific social goals, such as helping the middle class.

The new AI uses reinforcement learning to help policymakers predict the outcomes of different tax policies. The new framework uses AI agents, in this case simulated humans, to research how real people would react to various taxes. To make the simulation as realistic as possible, each AI agent earns money by collecting and trading resources and building houses. By adjusting their movement, trade and building behaviour, the AI agents learn to become more useful. During this simulation, the AI Economist learns to handle taxes and subsidies.

“Our model gives economists and policy-makers additional tools on which they can base their decisions,” Lead Research Scientist Nikhil Naik said in a Salesforce blog post. “By simulating millions of years of economies and finding a variety of tax frameworks, the AI Economist can predict how people would respond to a tax, like whether it will incentivise them to work more or work less.” 

Economic theory vs. AI

Salesforce research is based on the premise that the right tax policy can strike an optimal balance between productivity and equality. Some policies can increase both productivity and equality, but compromises are often made in a market economy. The use of economic models to predict the results of fiscal policy proved difficult for the researchers. In a blog post, they wrote that they did not take into account the fact the economy is not stagnant. Economic policies are continually changing, and employees are constantly acquiring new skills.

“Economic theory cannot fully mimic the complexity of the real world,” they wrote. “Instead, tax theory relies on simplifying assumptions that are difficult to validate. Moreover, experimenting with taxes in the real world is almost impossible.” According to Salesforce, the use of artificial intelligence can overcome some of these challenges.

The AI Economist is not perfect

It is important not to rely on new AI forms blindly. AI often lacks the small nuances that make the human life complex. The Salesforce research team also recognizes that “AI-based economic simulations still have limitations.” Nevertheless, they believe that the AI Economist can provide an excellent viewpoint of the financial possibilities in the real world.

“AI-based economic simulations do not yet model human-behavioural factors and interactions between people, including social considerations, and they consider a relatively small economy,” the blog post said. “However, these kinds of simulations provide a transparent and objective view on the economic consequences of different tax policies.”