Accountability and Trust Must Be Built Into the System
Ethical issues in artificial intelligence (AI) exist because rules of the road are needed to guide what is appropriate behavior in processing data against a set of algorithms to make decisions about individuals, communities and a society. Algorithms cannot think for themselves and are, therefore, morally neutral. A framework is needed to ensure they are designed in an ethical way. Fairness, trust, transparency, equality and accuracy have been identified as part of the ethical considerations that should be integrated into data analytics and AI.
Algorithms and Bots
Algorithms are a set of instructions designed to perform a certain task. It takes an input, analyzes the data, and produces an output. Airlines use algorithms to set passenger fares. The inputs include the class of service (economy, business, first class, etc.), the position of the seat in the aircraft (i.e., emergency exit row), the rate at which passengers are booking tickets, the time of the year and so on.
AI creates challenges such as eliminating bias from the system, building trust through transparency, protecting the privacy of data, and developing measures to ensure accountability. If the making of student loans is the goal, an algorithm can be developed to input data from prior student loans and establish a system to determine the ideal characteristics of a loan that is likely to be repaid. However, these characteristics may change over time so the system may become biased or one or more characteristics omits a portion of the population say, for example, minorities.
A bot (short for “robot”) is an automated program that runs over the internet. It is software that is designed to automate tasks you would do on your own, like making a dinner reservation and adding a doctor’s appointment to your calendar. Some bots are used to handle a variety of customer service requests, which would normally require a telephone call to a human agent.
Governance and accountability issues relate to who creates the ethical standards for the system, who governs the system, what are the internal controls over the data produced by the system, and who is accountable when unethical practices are identified. Here, the internal auditors have an important role to play.
Fraud detection is one of the biggest benefits of using AI in auditing. The system can be designed to catch anomalies. For example, an employee might submit paperwork for the reimbursement for a “meal” at a restaurant for $100. In a non-automated system, the request may be processed without any questions. In AI, the system may be set up to question even amounts of money. Upon further review, it may be determined that the $100 was for a gift certificate.
AI systems can have predictive value through machine learning and identifying high risk areas and events. It can devise an accounting fraud prediction model that more accurately calculates the probability of future material misstatements and improve the quality of audits.
A Deloitte survey of executives named ethical risks as one of the top three concerns related to AI. Ethical risks include how data is collected and processed; the validity of conclusions reached from the analysis of data, and the reproducibility in testing for consistent results and auditability.
There is somewhat of a global consensus about what should be the ethical principles in AI. They tend to include values such as transparency, non-maleficence, justice, responsibility and privacy. A brief discussion follows.
Transparency. The ability to understand the decisions of AI.
Non-maleficence. Never cause foreseeable or unintentional harm using AI, including discrimination, violation of privacy, or bodily harm.
Justice. Monitor AI to prevent or reduce bias.
Responsibility. Those involved in developing AI systems should be held accountable for their work.
Privacy. An ethical AI system promotes privacy both as a value to uphold and a right to be protected.
AI and Ethics in the Workplace
A workplace survey by Genesys examines the views of employers and employees with respect to AI ethics policies, potential misuse, liability and regulation. More than half of the employers questioned say their companies do not currently have a written policy on the ethical use of AI or bots and 21 percent expressed a concern that companies could use AI in an unethical manner. The research shows that both employers and employees support increased use of AI-enabled technologies in the workplace. Millennials are most apprehensive about the ethical use of AI. They worry about liability related to AI misuse and unethical uses of AI-produced data.
AI can help increase productivity, extend product life and minimize waste to make people’s lives better. But it’s only a tool that should be monitored to clearly identify what should and should not be done through the use of AI. This is where the ethical principles come into play.
As with any strategic initiative such as AI, the process should be guided by ethical standards. What are the right ways to use AI? How can AI be used to foster fairness, justice and transparency? What are the implications of using AI for productivity and performance evaluation? These are just a few of the questions to be answered to ensure ethical systems provide the foundation for the use of AI.
The next step is for the business community to work with government agencies to identify ethical principles in AI. Unfortunately, it seems the process is moving slowly and needs a nudge by technology companies, most of whom are directly affected by the ethical use of AI.
Blog posted on January 30, 2020 by Dr. Steven Mintz, who is a professor emeritus from Cal Poly San Luis Obispo. Steve recently published a book on personal development and wellness, “Beyond Happiness and Meaning: Transforming Your Life Through Ethical Behavior,” that is available on Amazon and Barnes & Noble. His Workplace Ethics Advice blog was named as one of the 30 Exceptional CSR Blogs. You can learn more about him at: https://www.stevenmintzethics.com/. You can follow him on social media at: https://www.facebook.com/StevenMintzEthics.