#154 Building Ethical Machines with Reid Blackman, Founder & CEO at Virtue Consultants
Sep 11, 2023
auto_awesome
Reid Blackman, Founder and CEO of Virtue Consultants and author of "Ethical Machines," dives into the complex world of AI ethics. He discusses the critical need for tailored approaches to mitigate bias and ensure fairness, using Amazon's biased resume screening as a case study. Blackman emphasizes accountability in AI decision-making and the importance of multidisciplinary teams. He explores data privacy concerns, the balance between transparency, and the ethical dilemmas of individual data control, urging a proactive and cohesive strategy for ethical AI integration.
Generative AI brings new risks such as manipulative conversational agents and privacy violations.
Organizations should be proactive in identifying and addressing biases in AI systems.
Senior leaders should take responsibility for implementing AI ethical risk programs and include expertise beyond data science.
Deep dives
Concerns in adopting AI for organizations
Organizations face concerns when trying to adopt AI, with the main concerns being ethically discriminatory or biased AI, privacy violations, and black box models. These risks have been well-established and discussed. However, the advent of generative AI has introduced new concerns, such as conversational agents being manipulative, IP and privacy violations, and problems with hallucinations. While these are the main concerns for corporations, there are other concerns like existential threats, job loss due to automation, and the spread of disinformation for society as a whole. It is important for organizations to identify and address these concerns in a responsible manner.
Misuses and failures in AI implementation
Misuses and failures in AI implementation can have significant consequences. An example of potential bias and misuse is the case of Amazon's resume reading AI, which was found to be biased against women. However, it is worth noting that Amazon took responsible action by recognizing the bias, attempting to mitigate it, and ultimately discontinuing the project when the bias could not be sufficiently addressed. This case highlights the importance of being proactive in identifying and addressing biases in AI systems. It also underscores the ease with which biased or discriminatory AI can be accidentally created and the challenges in finding effective mitigation strategies.
Accountability, fairness, and transparency in AI
Accountability is a critical aspect of AI deployment, and senior level leaders in organizations should take responsibility for implementing AI ethical risk programs. Data scientists and other professionals should be empowered to engage in ethical risk due diligence, but it is important to recognize that addressing ethical risks requires expertise beyond data science. Legal, ethical, and other perspectives are necessary to ensure a comprehensive approach. Transparency and explainability in AI decisions are also important, although there can be trade-offs between explainability and transparency. In high-stakes cases, such as the criminal justice system, the fairness of the procedure becomes crucial, and black box models may hinder the assessment of fairness. It is vital for organizations to carefully consider the ethical implications of AI implementation and take appropriate steps to mitigate biases, ensure fairness, and be transparent and accountable in their decision-making processes.
Adding to Existing Governance for Responsible AI Implementation
When implementing responsible AI, organizations should add to existing governance structures in a way that aligns with the organization's priorities. This could involve augmenting legal teams with ethicists, leveraging compliance boards, or providing additional training. Each organization's approach will vary based on its unique risk appetite and structure. It is important to avoid disruptive changes and instead integrate responsible AI practices seamlessly into existing processes.
Identifying High-Risk AI Use Cases and the Need for Ethical Considerations
Organizations need to consider the potential impact of AI use cases on people's access to basic necessities and human rights. Use cases related to jobs, credit or lending, healthcare, and life sciences are examples of potentially high-risk situations. Violating human rights or obstructing access to basic goods of life should be red flags. It is crucial for organizations to proactively evaluate the ethical implications of AI and determine when it is necessary to assemble an AI ethics committee or seek additional guidance.
It's been a year since ChatGPT burst onto the scene. It has given many of us a sense of the power and potential that LLMs hold in revolutionizing the global economy. But the power that generative AI brings also comes with inherent risks that need to be mitigated.
For those working in AI, the task at hand is monumental: to chart a safe and ethical course for the deployment and use of artificial intelligence. This isn't just a challenge; it's potentially one of the most important collective efforts of this decade. The stakes are high, involving not just technical and business considerations, but ethical and societal ones as well.
How do we ensure that AI systems are designed responsibly? How do we mitigate risks such as bias, privacy violations, and the potential for misuse? How do we assemble the right multidisciplinary mindset and expertise for addressing AI safety?
Reid Blackman, Ph.D., is the author of “Ethical Machines” (Harvard Business Review Press), creator and host of the podcast “Ethical Machines,” and Founder and CEO of Virtue, a digital ethical risk consultancy. He is also an advisor to the Canadian government on their federal AI regulations, was a founding member of EY’s AI Advisory Board, and a Senior Advisor to the Deloitte AI Institute. His work, which includes advising and speaking to organizations including AWS, US Bank, the FBI, NASA, and the World Economic Forum, has been profiled by The Wall Street Journal, the BBC, and Forbes. His written work appears in The Harvard Business Review and The New York Times. Prior to founding Virtue, Reid was a professor of philosophy at Colgate University and UNC-Chapel Hill.
In the episode, Reid and Richie discuss the dominant concerns in AI ethics, from biased AI and privacy violations to the challenges introduced by generative AI, such as manipulative agents and IP issues. They delve into the existential threats posed by AI, including shifts in the job market and disinformation. Reid also shares examples where unethical AI has led to AI projects being scrapped, the difficulty in mitigating bias, preemptive measures for ethical AI and much more.
