4 min read

Charting the Course: Navigating the ethical implementation of AI

Listen to this blog post instead:
Charting the Course: Navigating the ethical implementation of AI
Charting the Course: Navigating the ethical implementation of AI
7:18

In an era where AI is reshaping the very fabric of our digital world, as a Digital Leader you’re not only a pioneer of technology advancement but a steward of ethical responsibility.  

 

As this revolution weaves its way into our digital strategies, you’ll find yourself at a critical juncture: Does your own conscience align with your business conscience to implement AI in an ethical way? 

The ethics in the use, democratisation and monitoring of AI remain a hot topic, with the new UK AI Safety Institute, Centre for Data Ethics and Innovation (CDEI), AI Council and AI Procurement Guidelines working towards governing safe AI practices.   

As a Senior Project Manager with over 30 years' experience, in this blog I delve into the complex realm of Ethical AI implementation with a Conscience. You can also find helpful guidance on ethical considerations of AI implementation in our white paper.  

Where do you start when using AI ethically in Project Management? 

An Ethical AI Framework: Establish an Ethics Review Board:

Before commencing a project, I’m used to setting project standards, which for example in PRINCE2, encompasses a Project Board. For AI, this would include a dedicated Ethics Review Board of experts who can critically assess AI initiatives, particularly those with sensitive or high-stakes implications. Schedule regular board calls or meetings for the lifecycle of the project. This board serves as a gatekeeper for ethical considerations. 

Conduct ethical risk assessments:

I’ve found that the key to any successful project is to establish the risks as early as possible. As part of the project kick-off, risk workshops go beyond technical evaluations and dive into the social, economic, and political ramifications of AI applications. Conduct a comprehensive assessment of potential consequences and develop strategies to proactively address any ethical challenges. This isn’t a one-off assessment; you’ll need to monitor and manage the risks throughout the project lifecycle. 

Regular ethical audits:

Set up a regular schedule for ethical audits of your AI systems. Create a standard format for documenting results for consistency of reporting.  The audit should involve thoroughly examining AI performance, and its impact on various demographic groups. Essential is an iterative process of monitoring and improvement to align with evolving ethical standards and emerging technologies. 

Craft a crisis response plan:

In my various roles, I’ve found over the years that you can never prepare enough for what might go wrong! Prepare for potential ethical AI contingencies by establishing a clear Crisis Response Plan. Define roles and responsibilities to address issues such as data breaches, biases, or misuse.  Engage with those individuals to develop possible scenarios and responses ensuring close collaboration with the Data Protection Officer for the organisation. This will ensure a swift and lawful response to any ethical challenges that may arise during the project or once the system is deployed. 

Cultivate Ethical Excellence within your teams

Empower with training and awareness:

Elevate ethical awareness among your developers and data scientists by implementing targeted training programmes. They help equip team members to make ethically sound choices in their work, ensuring that ethical considerations are ingrained in every decision.  

Champion diversity and inclusion:

Form teams that embrace diversity and inclusion. Diverse perspectives are essential to address ethical challenges and mitigating biases in AI models. A wide range of voices and backgrounds can help uncover potential blind spots.  In my role, I encourage individuals to discuss blockers; the same applies to highlighting ethical challenges at daily stand-ups and at Sprint Review and Retrospectives. 

Prioritise transparency:

During development, emphasise transparent communication about AI systems. Encourage developers to clearly articulate how decisions are made and which data informs those decisions. Comprehensive documentation becomes a valuable ally in promoting transparency and accountability.  You should include documentation as a task to be completed when developing backlogs and sprints or you could include the availability of accurate documentation in acceptance criteria. 

Eradicate Bias and promote Fairness:

Take proactive steps to eliminate bias from AI algorithms. Conduct regular audits and include bias testing as part of the test process to identify and address potential biases. Implement bias-reduction techniques and fine-tune data to minimise any unjust bias, ensuring fairness at every step. Ethical AI demands resolute commitment to fairness and impartiality.  

Fuel your AI with data prowess

Governance is key

Just like in a Waterfall or Agile environment, establish and maintain governance practices. The same applies to robust data governance practices. Share this information with your stakeholders or Product Owner. Prioritise data privacy, security and adherence to critical regulations like GDPR. Strong governance serves as the bedrock of ethical data handling, instilling confidence in stakeholders and users.  

Streamline data collection

Optimise data collection practices by focusing on collecting only what's essential. Reducing data volume minimises the potential for misuse and breaches, enhancing your ethical data management reputation. 

Master data anonymisation

Safeguard privacy by anonymising data whenever possible. Data anonymisation acts as a shield, protecting individuals' privacy and reinforcing the ethical handling of data in AI projects. It's important to note that effective anonymisation depends on the specific use case for your organisation and the sensitivity of the data. Use a structured approach to help you make an informed decision.  Here are a few considerations I’ve learned as a Senior Project Manager:  

  • Assess data sensitivity 

  • Consider compliance requirements 

  • Understand how the data will be used 

  • Evaluate your budget and resource requirements to anonymise the data 

  • Determine the data retention period 

  • Evaluate the risk of re-identifying individuals from the anonymised data.

Lead the way!

As a Project Manager or Scrum Master, your conscience is to embody ethical values in every action and decision within your team. Be the vanguard for ethical AI practices across your team and the organisation, setting the standard that inspires others to follow your lead. 

To tackle ethical AI challenges effectively, adopt a holistic strategy that extends beyond just the technical considerations: embrace cultural, organisational and societal concerns. You must serve as an exemplary role model, with actions and decisions that consistently reflect ethical values.   

The journey to achieving this is by learning and understanding more about AI's complexities in a gradual, agile and experimental approach.

Our white paper provides you with further insight into the ethics and path forwards when  exploring the technology's many capabilities.