IBM Watsonx: An AI Value Creator for Accelerating Your Firm’s AI Governance (Part 2)
Previously, in our IBM Watsonx blog series, we explored the components within the IBM Watsonx and Watsonx.governance and the foundational aspects that make it a value-creator for exercising AI governance. In this blog, we examine IBM’s point of view on prioritizing AI ethics and what leaders should know about IBM’s commitment to building Responsible AI.
What IBM Wants Leaders to Know About Ethical AI Issues
AI tools have impacted millions with their large-scale automation and slew of intelligent capabilities. Even users with little to no technical expertise can leverage advanced versions of AI to achieve an organized reduction of repetitive tasks. According to Gartner, the AI software market will touch $139 billion over the next two years. However, the use of AI-based systems worldwide has raised several ethical reflections around gender and cultural stereotypes, biases, and concerns around risks to human rights, values, creativity, and originality.
These issues are not just our contemporary skeptics thinking out loud. In fact, these problems echo our society’s deep-rooted conditioning. Today, AI is behind in determining what appears in our search engine results. It drives decisions for public policy, talent hiring, loan approval processes, and so much more. Be it the unsolicited online product recommendations or the onslaught of prose and artistic drivel triggered by an input prompt, AI competes for tasks requiring human efforts and intelligence. So, there is a need for collective effort to re-examine the ethical implications arising in creating and large-scale deployment of emerging technologies within the realm of AI.
Amid the stage of Generative AI or GenAI adoption, where it has surpassed the peak of inflated expectations, many creators of AI tools are already under ethical and legal duress from global regulatory authorities and policymakers. There is an urgency to embed ethics in the design and development from the beginning of AI creation to inhibit inequities, harmful biases and discrimination, and threats to citizens’ privacy.
IBM is ahead of the curve with its proactive vision for AI ethics. It is one of the first tech firms in the industry to create an AI Ethics Board along with a company-wide AI ethics framework. In 2015, IBM hired its first AI ethics global leader and trained over a thousand ecosystem partners in 2022 on technology ethics. Recently, IBM was one of the eight companies to sign a voluntary commitment at the White House towards the development of safe and trustworthy AI. The commitment represents an opportunity to steer government action and Biden-Harris administration’s approach to managing the AI risks and promises and keeping a check on ethical issues with AI. Let’s dive into IBM’s core tenets for building ethical AI:
- Ethics Infused in Everyday Practices: IBM’s Ethics Board and Trust and Transparency Principles are the guiding force behind its transparent and ethical practices. The Ethics Board is made up of diverse talent within the company brought together to infuse principles and best-practices into the business and products. IBM’s five pillars for trustworthy AI—Explainability, Fairness, Robustness, Transparency, and Privacy underpin the development, deployment, design, and the use of AI systems:
- Explainability: The term Explainable AI or XAI denotes the level of transparency of AI systems at the algorithm recommendations level. So that it can be tracked for the model accuracy, fairness, and outcomes behind the AI-powered decisions. Explainability helps users and developers to make sure the AI system works as expected and comprehend and trust the results and the output produced by the ML algorithms, deep learning, and neural networks.
- Fairness: It refers to biases-free and equitable treatment of humans or a group of individuals by the AI systems. The overall intention is building AI systems that aid humans in making choices and decisions that are free of human biases to promote inclusivity.
- Robustness: AI systems and applications must be safeguarded against adversarial attacks and security risks. This would boost confidence in the reliability of the AI systems and their outcomes among the end-users.
- Transparency: Users of AI systems should have a deep level of understanding on how it works, evaluate their functionalities, and be aware of the merits and limitations.
- Privacy: Customers’ privacy and data rights are of utmost priority. AI systems must guarantee the highest levels of assurance to the users about the usage and protection of their personal data.
- Responsible AI: AI governance is the key to achieving Responsible AI (RAI). AI governance involves a set of processes of creating policies, delegating decision rights and ensuring organizational accountability for risks and investment decisions for the application and use of AI techniques. Many businesses rely on manual processes to track and document changes to data and AI model versions and are at high risk of errors and low transparency into the AI models, resulting in ‘Black Box’ approach that produces results that are unexplainable. IBM’s automated approach to governance with IBM Watsonx boosts an enterprise’s capability to adhere to regulatory requirements and address concerns and ethical issues with AI activities. IBM Watsonx.governance is built to manage and monitor the building, deployment, monitoring, and centralizing facts around AI explainability and transparency. The components within the platform are designed to manage the entire AI lifecycle, enabling governance for building Responsible AI.
- IBM’s Foundation Models: IBM Research’s Foundational Models for visual inspection are large scale deep-learning models that are pre-trained on non-annotated, domain specific datasets using self-supervision. They include billions of parameters that help enterprises build state-of-the-art models and repurpose them for implementing different use cases rather than training new models from scratch. This helps improve development of AI systems and pivot AI advancements in enterprises from the initial exploration to adoption phase. IBM’s foundation models are powered by the strength of the IBM research, product, and consulting teams, along with the innovations from the open global AI community and its secure, hybrid computing environment to build trustworthy AI.
- AI Factsheets: IBM has developed AI Factsheets to enable better AI governance and help deployers and users understand how an AI service or model was created. The Factsheets offer useful information on foundation models including training data performance metrics, biases evaluation, and energy consumption. They are modeled after the Supplier’s Declaration of Conformity used across industries for product and technical regulation conformity. Factsheets also provide documentation to AI developers on development best practices, risk management, data governance, technical documentation, record keeping, transparency, and cybersecurity of the foundation models. This document can be customized as per customers’ requirements and AI models to simplify compliance and enable effective AI governance.
Our Thoughts on Thwarting AI Ethical Issues with IBM Watsonx
As more organizations scale up the use of AI in their daily workflows for augmenting human efforts and capturing business efficiencies and profits, their leaders face tremendous pressure to uphold ethics. As a global community, we are aware of the splendid promise of AI in achieving success across humanitarian missions like finding cures for diseases, making vaccine discoveries, and even solving climate-related challenges. Concurrently, the world is also waking up to the cultural shift and evolving viewpoints on the fairness, transparency, and accountability of AI systems. The emerging focus on AI governance and growing familiarity of the concepts like Explainable AI and Responsible AI signify a beginning to era of conscientious innovations where ethics and morals are not an afterthought.
The future is not just about self-driven cars or breaking the linear perception about AI as a fully independent and error-free alternative to human efforts and intelligence. Understanding the ethical issues with AI is integral in building an informed community that rejects misinformation and prevents misuse of AI tools.
IBM Watsonx is reshaping the way enterprises approach their AI and data responsibilities for competitive differentiation and advance trustworthy AI. The general availability of IBM Watsonx.governance this December would help unlock opportunities to roll out effective AI governance strategies as a part of your organization’s governance, risk and compliance (GRC) goals. Learn how iTech GRC’s talented, certified team can help you rearchitect your approach towards building powerful governance and achieve transformations without encountering any ethical issues with AI.