Using Data and Respecting Users

Privacy and Security and Artificial Intelligence

Article Snapshot

Author(s)

Alisa Lenart and Marshall Van Alstyne

Source

Communications of the ACM, Vol. 63, No. 11, pp. 28-30, 2020

Summary

Firms should make ethical choices in using data to avoid souring relationships with users. Three basic guidelines reduce risk and help maintain user trust.

Policy Relevance

Users should benefit from firms’ use of their data. Detailed data should be discarded.

Main Points

  • In using data, firms should foster consumer trust by:
     
    • Securing users' data.
       
    • Offering reliable products and services.
       
    • Protecting users' legal rights, including intellectual property rights.
       
    • Acting ethically.
       
  • Firms can reduce risk by taking the customer's perspective, and by considering factors such as customer expectations and whether the customer receives fair value in exchange for her data.
     
  • Using customer data to fulfill a customer request is generally acceptable, as is using anonymized data for product improvement; sharing customer data with third parties for the third parties' benefit increases risk.
     
  • Risk remains low when data is used for the primary purpose for which the data was originally collected, but rises when the data is used for a secondary purpose.
     
  • Firms should design data collection systems that follow these three guidelines:
     
    • Data should be collected and processed in ways that benefit the user.
       
    • Only masked data should be saved.
       
    • After data is used to train algorithms, it should be discarded.
       
  • In designing for user benefit, the system should create more benefit than cost, the use of data should be ethical, and the primary beneficiary should be the individual who shared the data.
     
  • Firms should consider retaining only masked data; masked data, which cannot be traced back to its source, permits analysis while protecting privacy.
     
  • Data can be used to develop machine-learning algorithms to build a model of the world, and then be discarded.
     

Get The Article

Find the full article online

Search for Full Article

Share