## The Art in Artificial Intelligence: Make the Robots Serve the Public Good

_IMF Blog, January 11, 2018_

## Source details

**Canonical URL:** [The Art in Artificial Intelligence: Make the Robots Serve the Public Good](https://www.imf.org/en/blogs/articles/2018/01/11/the-art-in-artificial-intelligence-make-the-robots-serve-the-public-good)

## Other formats

- [Markdown version](/en/blogs/articles/2018/01/11/the-art-in-artificial-intelligence-make-the-robots-serve-the-public-good/index.md)
- [Structured JSON version](/en/blogs/articles/2018/01/11/the-art-in-artificial-intelligence-make-the-robots-serve-the-public-good/index.json)
- [Bundle manifest](/en/blogs/articles/2018/01/11/the-art-in-artificial-intelligence-make-the-robots-serve-the-public-good/bundle-manifest.json)

## Bibliographic details
- Authors: Brian McNeill
- Published: January 11, 2018

---

### Overview
- Author: Brian McNeill
- Date: January 11, 2018
- Core message: Artificial intelligence (AI) has rapidly matured and entered daily life; governments, policymakers, and the IMF must harness AI’s benefits and ensure AI serves the public good.

### Key areas of importance to the IMF
- Four areas of artificial intelligence and machine learning of importance to the IMF’s work:
  - Governance
  - Labor markets
  - Taxes
  - Social equity

### Governance
- Issues to address before basing analysis or policy advice on Big Data or algorithms:
  - Provenance of data.
  - Matters of privacy and informed consent.
- Characteristics of Big Data that complicate its use:
  - Dynamic and heterogeneous.
  - May originate in sectors that do not map cleanly to the IMF’s existing lines of responsibility or expertise.
  - Examples: data generated from e-commerce, the Internet of Things, satellite data, or supply-chain and logistics data.
- Institutional need:
  - Both the IMF and countries will need to develop expertise in the use of micro-level data.

### Labor markets
- Expected structural changes:
  - Fewer middle-skilled jobs (examples provided): insurance claims processing; jobs performed in a constrained physical space, like fork-lift operator or order expeditor.
  - These jobs have been more resistant to offshoring or automation so far but may disappear as AI improves and robots make decisions in ambiguous situations.
- Policy implications:
  - Impacts on education, retirement, and social welfare programs.
  - Potential for large numbers of middle-class jobs to be eliminated, leading to unemployment or underemployment.
  - Some jobs will require extensive retraining to ensure workers can perform new tasks.
  - Many countries are already facing rapidly aging populations; premature exit of workers from the labor market would make it more difficult for governments to fund social-welfare and retirement benefits.

### Taxes
- Fiscal challenge described:
  - If labor markets rapidly shed middle-skilled or low-skilled jobs, tax structures will need to reflect the decreasing share of GDP attributable to wages and salaries.
- Current revenue reliance cited:
  - Among the Organization for Economic Cooperation and Development countries, roughly half of government revenue is derived from individual income or social insurance taxes.
- Policy consideration:
  - Tax structures will need to change to sustain government revenues near current levels and to avoid creating further disincentives to job creation.
  - Example mentioned: Microsoft founder Bill Gates suggested that a tax might be levied on robots.

### Social equity
- Concerns about computer-driven decision-making:
  - Systems should be open to scrutiny and inspection.
  - Must not be automated versions of mental models that embed legacies of social inequality.
- Example risk:
  - Use of data for personalized pricing based on predictive models could lead to redlining of particular groups of customers and further marginalization, creating a self-fulfilling prophecy.

### Methodological and accountability challenges
- Contrast with economists’ practice:
  - Economists build models and refine them to reduce error and improve robustness.
- Challenge with many AI methods:
  - Some methods are impervious to external analysis because AI learns and adapts with new data; after millions of iterations, the algorithm can change substantially.
  - The explanation “The algorithm told me to do it,” is unlikely to withstand public inquiry as a basis for policy development.

### Next steps and IMF actions
- Institutional response:
  - The IMF will continue to bring in experts to promote information exchange.
  - The IMF will develop training so staff can work with emerging AI technologies.
- Intended outcome:
  - These activities will help the IMF work with its member countries to make artificial intelligence serve the public good.

*Source: The Art in Artificial Intelligence: Make the Robots Serve the Public Good — Brian McNeill, January 11, 2018*

---


_Source: https://www.imf.org/en/blogs/articles/2018/01/11/the-art-in-artificial-intelligence-make-the-robots-serve-the-public-good_