Tech Talk Interviews

Interviewer: How can organizations move beyond compliance to embed their ethics and values?

Rebecca: Compliance is what we must do. Ethics is what we should do. The first step is to stop copying frameworks from other companies and start tailoring your approach to your culture, operating model, and AI strategy. That tailoring includes defining your AI risk appetite, which clarifies where you will proceed, where you will pause, and what controls you require. There will always be areas of overlap across organizations, but the calibration needs to reflect your context.

Another key is building multi lens talent. If data scientists are creating models that affect people, they should pair strong technical judgment with a strong ethical lens. They need to understand how systems could impact individuals and communities, and how to mitigate those risks without losing sight of business objectives. Finally, lead with a sense of duty. Prioritise doing the right thing over short term gains, even when the latter looks attractive in a fast moving market.

Interviewer: What does implementing or incentivizing long term gains look like, and how do you ingrain that mindset?

Rebecca: Short term cost saving can be a business illusion if it erodes trust. If customer confidence or partner confidence is traded for quick wins, risk rises across legal, reputational, and operational dimensions. A timely example is the talent trap. It may look favorable on quarterly reports to reduce headcount while automating, but if you extract human expertise and historical knowledge, you weaken the very foundation you need to scale AI safely and sustainably.

There is also a lot of unproven technology right now. Tools that perform well in one context may not translate to another. Fit for purpose matters. Before you celebrate short term savings, test whether the solution aligns with your AI strategy, risk appetite, and operational realities. Incentives should reward resilience and reliability over time. That means funding robust evaluation, piloting with clear success criteria, and tying leadership goals to long term safety, quality, and performance metrics rather than only near term cost.

Interviewer: How important is it that a company first defines its values and what it stands for?

Rebecca: Most companies already have core values. Those values are the north star, and they existed before AI. The work now is to translate them into the AI landscape. What do our values mean in practice for our customers, our culture, and our employees as we deploy AI. Where could our systems affect rights, access, or opportunities. What level of documentation and oversight aligns with our standards. This is a deeper look at business values through an AI risk lens, turning high level principles into operating guidance that people can use when they build, buy, or implement AI systems.

Interviewer: How can companies demonstrate integrity toward customers and employees?

Rebecca: We need a balance of agility and strong foundations. At Weir, we emphasise a human centric framework with clear accountability across the enterprise. AI risk should be owned across the business, not isolated in technical or compliance teams. To make that real, you need robust structures and forums where collaboration is the norm, and where the right people are at the table at the right time to ask hard questions, provide guidance, identify risk, and also spot business value.

This only works if people are empowered. Invest in AI literacy programs so that product teams, legal teams, operations, HR, and leadership have the knowledge to make informed decisions. Integrity shows up in how you govern data, how you evaluate vendors, how you monitor models, and how you communicate limitations and recourse to users. Make these practices visible and consistent, inside and outside the organisation.

Interviewer: What are some of the right questions that should be asked?

Rebecca: Start at the strategic level. Is this AI solution aligned with our AI strategy and business goals. What specific value will it deliver, and what risks does it introduce. Are the risks within our appetite. Are risks proportionately controlled with the right mix of technical and procedural safeguards. Do we have the right human oversight for this use case. Who is accountable for outcomes and what is the escalation path if things go wrong.

To answer these questions well, you need shared knowledge. That includes understanding the data the model uses, the populations it will impact, the performance it has demonstrated under stress, and the failure modes you can anticipate. Good questions are only as strong as the evidence you bring to them.

Interviewer: What steps can organisations take to improve transparency and readiness?

Rebecca: As AI becomes more democratised, we need to simplify without dumbing down. Translate complex data science concepts into human readable information. Model summaries similar to nutrition labels are useful, capturing risks, potential bias, drift tendencies, and known limitations in a clear format. If someone in HR is procuring a third party solution, they need to see those details up front so they can calibrate the level of human oversight and understand where the tool should not be used.

Context is king. A model validated in one setting may behave differently in another. An industrial system that performs well in a factory in Detroit may expose different risks in a factory in London. Readiness means testing for your purpose, your environment, your data, and your users. Build processes for due diligence, sandbox testing, and phased rollout, then monitor in production with clear thresholds and stop conditions.

Interviewer: AI is evolving quickly and the key trends seem to change weekly. What do you see as the key trends shaping AI now?

Rebecca: We have moved from the wow phase to the how phase. Many teams are advancing from assistance to automated agents that can take action across systems. That shift increases responsibility. We need strong governance, clear oversight, and the right people, processes, and tools to deploy safely at scale. Collaboration is essential because no single function can see the entire risk and value picture.

AI literacy remains a foundational trend. Empowered teams make better choices about data, architecture, evaluation, and incident response. With that foundation, you can scale solutions that create business value while staying within your safety and sustainability commitments. The long game is trustworthy automation that remains manageable over time, not just impressive demonstrations.

Conclusion:

Embedding ethics in AI is a practical discipline rooted in clarity, accountability, and education. When organisations align strategy, talent, and governance, they can innovate with confidence and earn durable trust.