When using AI to support business-critical decisions, it's important to ensure decisions are being made responsibly. This means, in part, without breaching data privacy policies, or introducing racial, gender or any other bias. Companies must take on the challenge of making sure that the AI acts in a responsible way. Key questions include:
Are accurate, bias-free decisions being made?
Is anyone's privacy being violated?
Is the technology being appropriately governed and monitored?
Responsible AI (RAI) is artificial intelligence that is lawful and that adheres to well-defined ethical guidelines regarding fundamental values, including individual rights, privacy, non-discrimination, and non-manipulation. RAI principles are increasingly important with the rise of new generative AI models and upcoming regulation in the sector. RAI can also be a competitive advantage because consumers have growing concerns about privacy and biassed decisions.
At PwC Belgium, we’ve been supporting companies of all sizes and in all sectors with their AI-related projects for years. We’ve encountered all the challenges and risks that AI can bring and have developed solutions for each of them. To implement responsible AI, we’ve compiled all our experiences, lessons learned, and best practices into ready-to-use tools tailored to your company's needs.
Are you ready to implement responsible AI? Take our Responsible AI Diagnostic test to find out!
On 21 April 2021, the European Commission released its much-awaited first draft of the Artificial Intelligence Act. The proposed legal framework focuses on the use of AI systems and the associated risks. It proposes to create a technology-neutral definition of AI systems in EU law, and to set out classifications for AI systems with different requirements and obligations according to a ‘risk-based approach’. Last December, the EU Council adopted its common position on the AI act. Now, the European Parliament has to discuss and adopt its own position before the final negotiations starts.
In 2018, the European Commission outlined its vision for artificial intelligence, which promotes ethical, secure, and innovative AI created in Europe.
In 2019, the European Commission established a framework for creating trustworthy AI, which includes seven essential requirements.
In 2021, the European Commission unveiled a new proposal for a legal framework focused on the specific use of AI systems and associated risks.
In December 2022, The Council adopted its general approach on the EU AI Act. The text clarifies key concepts of the upcoming regulation and sets criteria to define AI systems.
In March 2023, the European Parliament finalised his own position about the general proposal. The trilogues between the Parliament, the Council and the Commission will start. By the end of 2023, the AI Act will be adopted.
During 2024, policy makers will be working on harmonised versions of the standards and will be implementing the acts.
In 2025, the transitional period will end. The AI Act will then become applicable.
Some AI systems presenting ‘unacceptable’ risks would be prohibited, for example AI systems used for social scoring. A wide range of ‘high risk’ AI systems would be authorised, but subject to a set of requirements and obligations to gain access to the EU market. AI-based cameras for autonomous vehicles will typically fall into this category. Those AI systems presenting only ‘limited risk’ would be subject to very light transparency obligations. That would be the case for chatbots or spam filters.
Our Responsible AI Toolkit is a customizable suite of frameworks, tools, and processes designed to help you use AI ethically and responsibly, from strategy to execution. We'll tailor our solutions to your organisation's unique business requirements and AI maturity level.
Who is accountable for your AI system?
Is your AI unbiased? Is it fair?
How was that decision made?
Will your AI behave as intended? What are the security risks and implications that should be managed?
Is your AI legal and ethical?
Responsible AI is of utmost importance for companies of all sizes, regardless of the complexity of their AI systems. It is recommended that businesses just beginning to develop AI algorithms incorporate responsible practices into their systems, processes, and corporate culture from the start. This is because it is often easier to address issues such as security, compliance, and bias in the early stages of production, rather than when the system is already established and running. It’s advisable to begin integrating responsible practices into your organisation now. By doing so, you can ensure that the AI you are developing is effective, ethical and responsible.
By improving your models' transparency and explainability you can make more informed business decisions with trust and become more competitive in the market. This approach can also help you to foster better relationships with stakeholders, build customer trust, attract and retain talent, comply with regulations, manage risks more effectively, and boost revenue.
Whether you're just starting out or getting ready to scale, responsible AI can help. Drawing on our expertise in AI innovation and deep global business knowledge, we'll evaluate your end-to-end needs, and design a solution to help you address your unique risks and challenges.
We have the right team to build AI responsibly, both internally and for our clients. We can bring big ideas to life at all stages of AI adoption.
And with our Responsible AI Toolkit, we bring a collection of frameworks, templates, and code-based assets to streamline your RAI journey. These accelerators include:
AI and Data Ethics Traceability Matrix
Ethical AI and Data Framework
AI and Data Ethics Impact Assessment and Organisational Maturity Assessment
Data Protection and AI Risk and Controls Library
Project Management
AI and Data Ethics Training
The development of rigorous frameworks to shape decision-making in public sector organisations will be crucial for realising AI’s potential to transform public services and administration. Ethical decisions regarding citizens’ well-being must be at the forefront of the government's efforts to explore and adopt this technology. The public sector can be a leader in adopting secure, trustworthy and sustainable AI systems. Europe's public sector has the potential to leverage its significant collective purchasing power to act as a catalyst and stimulate demand for responsible AI.
As an AI expert, PwC has already collaborated with public organisations such as the Federal Public Service Economy and the Brussels region to develop their AI action plans. Responsible AI has been systematically addressed to ensure it is considered when implementing AI in Belgium.
AI in the financial sector has been a game-changer, automating processes to make them more efficient and cost-effective. Credit adjudication, personalised marketing offers, next-best action, cybersecurity, and fraud detection are all examples of AI applications. The AI Act will ensure that societal and environmental well-being, diversity, non-discrimination, fairness, and transparency are taken into account in these use cases. Financial firms must consider how compliance can be achieved and how the new rules will interact with other existing (such as GDPR, MiFID II or RTS 6).
To that end, PwC has already supported European banks and insurance companies in auditing AI systems and in creating a robust framework to proactively identify risks and the appropriate controls to demonstrate ongoing compliance.
AI has the potential to revolutionise healthcare, including, but not limited to, clinical practice, biomedical research, public health, and health administration. Responsible AI is especially important to reduce the risks associated with AI, such as patient harm caused by AI mistakes, misuse of medical AI tools, bias, lack of transparency, and privacy and security issues. Existing regulations already set out detailed requirements for medical device safety and performance. The AI Act will bring additional regulatory obligations and ethical considerations, however, it should not impede the adoption of AI, which holds immense potential for the healthcare industry.
PwC has a long history of collaboration with healthcare providers. We’ve helped large pharmaceutical companies to implement data ethics and develop strategic data and AI ethical guidelines.
