资源描述
Why addressing ethical questions in AI will benet organizationsWhy addressing ethical questions in AI will benefit organizations Executive Summary Every wave of technology development raises its own set of ethical questions, from unintended and harmful consequences for users to concerns about technologies being weaponized. The pace of technological advancement tends to outstrip the pace of regulatory and ethical frameworks a position that AI finds itself in today. AI has unleashed a range of ethical questions, from concerns over autonomous vehicles to what constitutes end-user consent. To probe these issues, we have conducted a comprehensive survey of both business executives and individuals to understand ethics and the transparency of AI-enabled interactions and what organizations are doing to address any concerns. We surveyed 1,580 executives in 510 organizations and over 4,400 consumers across countries such as the US, UK, China, Germany, and France. We found that: Getting ethics in AI right will benefit organizations and mitigate risks AI interactions that consumers and citizens perceive as ethical build trust and satisfaction: When consumers and citizens feel an organization offers ethical AI interactions, over half said that they would place higher trust in it, share their positive experience, be more loyal, purchase more, and advocate for it. Organizations whose AI systems consumers see as interacting ethically enjoy a 44-point Net Promoter Score (NPS) #advantage compared to those seen as not using AI ethically. AI interactions perceived as unethical can damage brand reputation: Nearly two in five consumers would complain to the company and demand an explanation if they experienced an unethical interaction. In the worst case, a third would stop interacting with that company. Most organizations have encountered ethical issues in AI over the last two to three years. Consumers seem to corroborate this finding Most executives (77%) are uncertain about the ethics and the transparency of their AI systems Executives in nine out of ten organizations are aware of at least one instance of the use of AI systems that resulted in ethical issues (see insert below for examples of ethical issues) Close to half of consumers believe they have felt the impact of an ethical issue (to probe this matter, we gave survey respondents specific instances of unethical practices detailed in the appendix at the end of the report). This is leading a majority of consumers (76%) to advocate regulations on the use of AI: they expect governments, independent bodies (e.g. IEEE) and regulators to lay down the principles that define the just use of AI The pressure to implement AI is fueling ethical issues: Executives identify the pressure to urgently implement AI as the top reason why ethical issues arise from the use of AI. Executives are starting to realize the importance of ethical AI and are taking action when ethical issues are raised 51% of executives believe that it is important to ensure that AI systems are ethical and transparent 41% of senior executives report that they have abandoned an AI system altogether when ethics concerns were raised; 55% implemented a “watered - down” version of the system. 2# Net Promoter, NPS , and the NPS -related emoticons are registered service marks, and Net Promoter Score and Net Promoter System are service marks, of Bain & Company, Inc., Satmetrix Systems, Inc. and Fred Reichheld. What kinds of ethical issues emerge from the use of AI? By ethical issues arising from the use of AI, we mean the interactions that result in outcomes that are unexplainable, unfair, not transparent and/or biased against a certain group of users. Examples of AI interactions that may result in ethical issues include: Using AI to screen job applicants that results in a disproportionate selection of candidates across gender, ethnicity, age, or other factors Using AI for surveillance at the workplace without the consent of employees Using an AI system that is not able to explain how it arrived at a decision to deny credit or an insurance claim or cannot be audited on its decision making Collecting and processing healthcare personal data in AI algorithms without consent Discriminatory pricing or availability of products/ services resulting from automated AI decisions. First steps to proactively address ethics in AI Our previous research has established that AI has significant potential across industries and functions. 1By addressing ethics issues upfront, we believe that organizations stand to gain additional benefits as well as avoid regulatory, legal, and financial risks that may result from a market or public backlash on AI. We have developed a comprehensive approach to help organizations proactively address ethics questions in AI. This approach targets three key groups in the organization: General management (CXOs, and business leaders and those who have questions of ethics and trust in their remit), who will be responsible for laying the foundational practices and processes for ethical AI Customer-and employee-facing teams (e.g., HR, marketing, communications, and customer service) that are responsible for deploying AI ethically for users, and AI, data, operations, and IT teams that design, develop, and implement AI systems For CXOs, business leaders and those with a remit for trust and ethics Lay down a strong foundation with a strategy and code of conduct for ethical AI Develop policies that define acceptable practices for the workforce Build awareness of ethical issues across the organization Create ethics governance structures and ensure accountability for AI systems Build diverse teams to inculcate sensitivity to ethical issues For customer-and employee-facing teams such as HR and marketing For AI, data, and IT teams Ensure that ethics questions are debated from the design of AI-infused applications by clearly defining the outcomes and intent of these applications, and ensuring accountability for them Educate and inform users on what building trust in AI applications means Empower users with more control and ability to seek recourse Making AI systems transparent and understandable Practice good data management in compliance with regulations where they exist, and address and monitor potential biases in data Use technology tools to actually build ethics in AI (bias detection, transparency, explainability, continuous monitoring of accuracy) 3Artificial intelligence offers a huge opportunity for businesses and the economy, but significant questions are being raised about the ethical issues surrounding this technology. To examine these questions more closely, we have undertaken this research to understand the current relationship between AI applications and their users. As Figure 1 shows, we examined both sides of the debate from a business and end-user perspective. Introduction 4 Why addressing ethical questions in AI will benefit organizations 4 Why addressing ethical questions in AI will benefit organizations
展开阅读全文