Rohit Singh London in Artificial Intelligence, IT - Information Technology, Business Director - Head of Innovation & Digital Disruption • KPMG Apr 2, 2019 · 1 min read · +800

Will Predictive Analysis Save Humanity or Destroy It?

Will Predictive Analysis Save Humanity or Destroy It?

The simple answer: it depends. At least that’s the answer given by Dennis Hirsch, a law professor at Ohio State and head of Ohio State’s Program on Data and Governance. In a recent interview, he outlined ways to regulate predictive analytics for society’s benefit.

Predictive Scale

Data sets have grown incredibly large. This makes it possible for marketers to infer future consumer behavior from consumer’s previous purchasing patterns. Yet, consumers have not directly consented to provide purchasing data for the purpose of predictive analytics. This is the part of the scenario where ethical questions arise.

Consumers have not granted marketers, or anyone else, direct consent to make inferences based on their data. Companies can use inferences to inform algorithms. These algorithms then have the potential to cause bias or to embed procedural unfairness into otherwise objective systems.

The Power of Inference

This might not seem to matter, but inference has power. Risk analysis tools are designed specifically to infer risk. They can be used during loan acquisition, house purchases and even job application processes. Any situation in which a company is at risk triggers a need for predictive analysis.

What began as a marketing tool has become a powerful underwriting decision-maker. Since predictive analytics is still in its early stages of implementation, the need for ethical intervention is pertinent.

Ethical Considerations

Hirsch explains that predictive analytics are helpful. They make it simple to get a quote on an insurance rate, for example. They also assist in small things, like getting a hint about the next movie you might want to watch.

But when left unregulated, even the most benign inferences can impact decisions. Without regulation, decisions like who gets to buy a house, who gets to post bail or who gets an offer for a free flu shot can be biased.

When it comes to ethics, Hirsch recommends establishing an agency to evaluate how companies are using data. He suggests using these four factors:

  • Privacy
  • Manipulation
  • Bias
  • Procedural Fairness

Hirsch’s suggestion seems reasonable. A broader look at predictive analytics suggest reason might be lost when applied to the emerging data economy. As China, the United States and other nations amp up their data collection prowess, global data may not lend itself to any regulatory authority. In this scenario, humanity won’t be destroyed but humanity should proceed with caution.