Data In, Data Out: The Hidden Truths Behind AI, Bias and Trust

Across Brisbane, Sydney and Melbourne, our Data In, Data Out panel brought together leaders, technologists and decision-makers to unpack a simple but powerful truth:

What goes in absolutely determines what comes out.

As AI continues to influence the way we collect and analyse data, strategy is only as powerful as the integrity behind the data that fuels it. And while we often focus on tools, automation and acceleration, this session was designed to spark the uncomfortable, but necessary, conversations about what sits beneath the surface.

Because as we dive deeper into an algorithm-driven future, one thing became clear… Decisions are only as fair as the data behind them.

Why This Discussion Matters Right Now 

An excerpt from Carrie Mott’s Blog ‘Women In Digital “Data In, Data Out” Series: When AI Moves Fast, Trust Has to Move Faster’

This “Data In, Data Out” series was designed to be practical, honest and grounded in lived experience. Not theory. Not hype. The real trade-offs. The blind spots. The uncomfortable “wait… are we sure this dataset should even exist?” moments. 

We are seeing a widening gap between ambition and readiness.

ADAPT’s State of the Nation: Data and AI in Australia 2025 shows only 24% of leaders believe their data is AI-ready, despite AI being treated as strategic across many organisations. That gap is not abstract. It is structural. And it is growing.

Globally, McKinsey reports that 65% of organisations are now using generative AI regularly in at least one business function, nearly double the previous year. Yet trust in digital systems continues to fluctuate as high-profile data incidents persist. Ambition is accelerating. Foundations are not always keeping pace. And the broader context matters.

UN Women Australia’s International Women’s Day 2026 theme, “Balance the Scales”, alongside the global campaign “Give to Gain”, calls for structural fairness, shared accountability and collective action. If we want fair outcomes at scale, we have to build fair systems from the start. Which brings us back to data.

Key Takeaways from the Panel

Bias is not just a data problem

We often blame the algorithm. But one of the strongest themes across all three cities was this: bias is most dangerous during interpretation and application.

You can have “clean” data. You can have technically robust models. But without context, diversity of thought and critical oversight, you will still fail.

Bias does not magically disappear at scale. If it exists in the input, it multiplies in the output. And in a world where AI consumes and generates information at unprecedented speed, that amplification happens almost instantly.

Traditionally, bias travelled at the speed of human reporting. Now, it moves at the speed of machines. Which makes Human-in-the-Loop (HITL) governance more critical than ever.

Trust Is the Ultimate Currency

Technology adoption is not just about capability. It is about belief. We discussed the real risk of failing data integrity: erosion of trust.

If a frontline team uses a “Next Best Action” tool and it fails them twice, they will stop using it. It doesn’t matter how sophisticated the model is behind the scenes. Cultural buy-in disappears quickly. And without trust, you are not a data-driven organisation, you are simply a data-producing one.

Trust is built through:

• Transparency
• Traceability
• Clear ownership
• Accountability in action

Diversity Is a Technical Requirement

Another powerful theme across the panels: diversity is a technical necessity. If the room where the models are built is not diverse, the insights will not be either. Reducing bias requires expanding perspective — across data science teams, leadership, product owners and decision-makers. Inclusion cannot be assumed. It must be designed.

The decentralisation of data tasks makes this even more important. Today, almost everyone manages data in some form — but not everyone is trained as a data manager.

AI may feel like the answer to life, the universe and everything, but the fundamentals still matter:

• Data quality
• Data completeness
• Context across the lifecycle
• Ethical interpretation

And layered across all of this is something deeply human: psychological bias. That psychological layer, unique to our lived experiences, influences how data is framed, questioned and acted upon.

The Simple Truth We Couldn’t Escape

Across three cities and countless conversations, we kept coming back to the same conclusion:

Data integrity is strategy.
Inclusion is architecture.
Trust is currency.

And what goes in absolutely determines what comes out.

The future of AI will not be defined by the tools we adopt, but by the standards we set – in leadership, governance, diversity and culture.

Thank you to everyone who joined us in Brisbane, Sydney and Melbourne for leaning into the complexity and engaging in the conversations that matter.

• Brisbane – Hosts, Cisco, Terry Weber and Carrie Mott, our MC Katrina Edgar and our panel of speakers, including Karyn Smith, Craig Dargusch and Olivia Segedin.
• Sydney – Hosts, Sekuro and Sandie Forrest, our MC Patricia Droguett and our panel of speakers, including Catherine Lee, Joseph Gurney and Varsha Nimalendran.
• Melbourne – Hosts IM Academy, our MC Su Jella and our panel of speakers, including Laura Parr, Sophia Duan and Kathryn Collier.  

And, of course, to everyone who attended and contributed to the conversation! Stay curious, question everything, and remember: What goes IN absolutely determines what comes OUT.