IBM's AI Ethics Board demonstrates a more comprehensive and transparent approach with publicly available resources, dedicated training, and clear reporting mechanisms. DeepMind's program, while robust, lacks some of the transparency and readily available details found in IBM's.
| Attribute | DeepMind's Ethics & Society Team | IBM's AI Ethics Board |
|---|---|---|
| Transparency of ethical guidelines | DeepMind has faced criticism regarding the transparency of its AI ethics board, with concerns raised about the lack of publicly available information on its members, discussions, and meetings. However, DeepMind has more recently committed to providing early access to its AI models for research into evaluation and safety. | IBM emphasizes transparency through disclosure, striving to make AI systems understandable and accountable. They are committed to educating people and society about how AI works. IBM's Principles for Trust and Transparency are publicly available and guide their approach to AI ethics. These principles are comprehensive, covering the purpose of AI, data ownership, and the need for transparency and explainability. |
| Scope of ethical considerations | DeepMind's ethical considerations encompass a broad spectrum of AI-related risks, including bias, privacy, safety, and security. They take a holistic approach to responsibility, safety, and security, guided by their AI Principles. | IBM's AI ethics framework addresses issues like data responsibility and privacy, fairness, explainability, robustness, transparency, environmental sustainability, inclusion, moral agency, value alignment, accountability, trust, and technology misuse. They also consider potential issues around mental privacy, human agency, and identity related to neurotechnology and the responsible use of quantum computing. IBM uses tools like the AI Fairness 360 Toolkit to detect and mitigate bias in machine learning models. They also emphasize diverse and representative data, ongoing bias detection, and human-in-the-loop processes. |
| Stakeholder engagement mechanisms | DeepMind engages with various communities, including educators, artists, and people with disabilities, to understand their needs and ensure AI is developed and deployed inclusively. They also collaborate with researchers across industry and academia, engaging with governments and civil society to address challenges. | IBM promotes a cross-functional, collaborative approach to AI ethics, involving all stakeholders in the AI ethics process. They also gather input from society about the ways AI is affecting the world. They also have an Advocacy Network, a grassroots-level network of employees who share and promote IBM's technology ethics principles. |
| Methods for auditing and accountability | DeepMind proposes a three-layered framework for evaluating the social and ethical risks of AI systems, including assessments of AI system capability, human interaction, and systemic impacts. They also conduct safety evaluations for bias and toxicity. | IBM emphasizes regular audits and monitoring of AI systems to identify and correct biases or errors. The AI Ethics Board is a mechanism by which IBM holds itself accountable. |
| Integration with product development lifecycle | Google DeepMind emphasizes designing AI to be safe from the beginning, rather than thinking about safety afterward. | IBM uses an 'Ethics by Design' framework to integrate tech ethics into the technology development pipeline, including AI systems. This involves embedding tech ethics principles throughout IBM's products, services, and broader operations. |
| Resources allocated to ethics program | DeepMind has world-class teams focusing on technical safety, ethics, governance, security, and public engagement. They also launched a new research unit, DeepMind Ethics & Society, to complement their work in AI science and application. | Specific budget details are not available in the search results. However, the existence of the AI Ethics Board, dedicated roles like AI Ethics Focal Points, and the CPO AI Ethics Project Office suggest a significant investment. |
| External collaborations and partnerships | DeepMind collaborates with various organizations, including academic institutions, non-profits, and other companies, to apply AI to solve global challenges and mitigate risks. They co-founded the Partnership on AI and the Frontier Model Forum. | IBM collaborates with academic institutions, NGOs, and other organizations. They are a founding member of the Partnership on AI and the Vatican's 'Rome Call for AI Ethics'. IBM and Meta co-founded the AI Alliance. |
| Impact on AI product design and deployment | DeepMind's AI ethics program aims to ensure that AI systems are respectful of human rights, equality, and civil rights laws. | The AI Ethics Board reviews use cases to ensure they are consistent with IBM's principles and core values. IBM's framework is designed to be operationalized throughout the AI lifecycle, from initial concept through deployment and monitoring. One example is the use of the AI Fairness 360 toolkit to assess lending models in a financial institution. |
| Employee training and awareness programs | Details on specific employee training programs are not readily available in the search results. | IBM provides role-based educational programs to raise awareness. All employees are being trained in the 'ethics by design' methodology. |
| Mechanisms for reporting and addressing ethical concerns | The search results do not provide specific details on mechanisms for reporting and addressing ethical concerns. | IBM has established AI Ethics Focal Points who act as points of first contact to identify concerns, mitigate related risks, and escalate issues to the AI Ethics Board when required. |
| Publicly available documentation and reports | Google provides AI Responsibility Reports. | IBM provides resources such as the IBM AI Ethics Webpage, Principles for Trust and Transparency, and Pillars of Trust. They also publish white papers and blog posts on AI ethics. |
| Influence on industry standards and best practices | DeepMind actively participates in defining best practices and setting safety and security benchmarks through organizations like MLCommons and the Frontier Model Forum. | IBM advocates for policies to hold AI to high ethical standards. They also contribute to the development of industry standards and best practices for AI ethics. |