The Role of Ethical Principles in AI Startups

Abstract

Do high-tech startups benefit from having more ethical artificial intelligence (AI) practices? Even without AI regulations, startups frequently adopt ethical AI policies, consult expert guidance, train employees about bias, and hire minority programmers to manage ethical issues associated with data collection, storage, and usage. In this paper, we use novel data to describe startups' ethics-related actions and test whether stakeholders (i.e., suppliers, customers, and investors) influence these ethics-related actions. We find that relationships with upstream technology suppliers, like Amazon, Google, or Microsoft, relate to increased ethical AI policy adoption. However, when we consider outcomes, merely adopting an ethical AI policy does not relate to increased fundraising performance. Instead, investors appear to reward startups that take more costly pro-ethics actions. Other stakeholder relationships, like a relationship with a venture capital investor or a founder with an MBA, relate to engaging in more pro-ethics actions. We argue that these patterns are consistent with existing theories, notably signaling and knowledge transfer

Keywords: data, ai startups, ethical ai policies, algorithmic bias, ethics, fairness, startups, entrepreneurship

JEL Classification: M13, O32, D82