In this episode of Siren Investigates, Siren CEO John Randles sits down with Julian Stodd, writer, researcher and founder of Sea Salt Learning, to explore how generative AI is reshaping the way we understand knowledge, trust, identity and security.
Julian is the author of 17 books and more than 2,000 articles, with his recent work exploring the impact of generative AI beyond individual technologies and tools. Rather than asking simply how organisations should use AI, he argues that leaders need to consider how AI is changing the systems in which they operate.
Their conversation explores how generative AI is challenging established ideas of truth and identity, what this means for law enforcement and security organisations, why uncertainty can be valuable, and how investigators can combine emerging technology with human judgement.
How is generative AI changing the way we understand truth, trust and identity?
Generative AI is changing more than the speed at which information can be created. Julian argues that it is challenging some of the structures people have traditionally relied upon to determine what is real, trustworthy and authentic.
The cost of producing convincing content has fallen dramatically. Written material, images and other forms of communication can increasingly be generated and replicated at enormous scale.
That presents a particular challenge for the mental filters people have traditionally used to judge authenticity.
An obvious phishing email containing poor grammar might once have been relatively easy to identify. Generative AI can produce communication that is significantly more convincing, making familiar signals of authenticity less dependable.
At the same time, our understanding of identity is becoming more complex.
Julian describes modern identity as fragmented and fluid. People can inhabit different communities, maintain different aspects of themselves across physical and digital environments and behave differently depending on the social context.
AI adds another dimension by making it increasingly easy to create or imitate identities at scale.
The result is a world in which organisations and individuals may need new ways of establishing trust rather than relying on assumptions that previously seemed dependable.
How can law enforcement and security organisations adapt to AI?
For law enforcement, defence and national security organisations, Julian argues that AI should not be treated simply as another technology added to an otherwise stable environment.
The wider system itself is changing.
Criminals, hostile actors and other groups can access many of the same emerging technologies as the organisations trying to investigate them. Generative AI can lower the cost of producing disinformation, creating identities and operating at scale.
Traditional security organisations face a different reality. They may be working with legacy technology, established procedures, finite budgets and regulatory requirements while responding to threats that can evolve much faster.
Julian argues that these organisations therefore need to determine what must remain protected and where they can create room for experimentation.
Core principles such as justice, fairness, security and democratic accountability may remain essential. The systems and methods used to protect them, however, may need to evolve.
In the short term, organisations can create structures that allow people to explore emerging technologies, question established assumptions and encounter different perspectives.
Over the longer term, the challenge becomes more fundamental: organisations may need to reconsider how they are structured to operate effectively in a world where technology, threats and social behaviour can change rapidly.
Why should organisations embrace uncertainty and ambiguity?
Organisations are traditionally designed to reward certainty.
Leaders are expected to know what is happening, explain what should happen next and make decisions with confidence. Julian argues that this can become a weakness when the environment itself is changing quickly.
Uncertainty can contain useful information.
Different opinions, weak signals and ideas that challenge an established interpretation may reveal something that a more rigid system misses.
Julian points to major organisational failures where information was available but the systems responsible for making decisions failed to give enough weight to dissenting voices.
Rather than attempting to eliminate ambiguity immediately, organisations can create space to examine it.
For investigative and security organisations, this is particularly relevant. Investigators ultimately need evidence and conclusions that can withstand scrutiny, but reaching those conclusions can require exploring competing hypotheses and questioning assumptions along the way.
Julian describes the challenge as maintaining a dynamic tension between certainty and exploration: protecting what must remain stable while creating enough freedom to test new ideas, learn from failure and adapt.
How can AI help investigators make better decisions?
Investigators are often required to reach conclusions from large amounts of incomplete, complex or conflicting information.
Julian suggests that one of the most useful technological capabilities would be something that helps an investigator understand their certainty within a broader context.
When people develop a strong theory about what happened, there is a risk that assumptions supporting that theory become increasingly difficult to see.
AI could potentially help challenge that process.
Rather than simply producing another answer, technology could help identify where an investigative conclusion is supported by strong evidence, where assumptions have been made, where ambiguity remains and where alternative interpretations deserve consideration.
Julian describes the idea as something resembling a “heat map of certainty.”
The objective would not be to prevent investigators from reaching conclusions. Investigations ultimately require outcomes.
Instead, technology could help investigators understand the path they have taken towards that conclusion and recognise where greater scrutiny may be required.
Combined with human judgement, this type of capability could help investigators test assumptions, consider alternative hypotheses and ultimately develop greater confidence in the conclusions they reach.
Key takeaways
- Generative AI is challenging established assumptions about truth, trust, identity and authenticity.
- The falling cost of generating convincing content creates new challenges for traditional methods of identifying misinformation and deception.
- Law enforcement and security organisations need to adapt not only their technology, but also how they respond to uncertainty and rapid change.
- Organisations should protect essential principles while creating space to experiment with new technologies and approaches.
- Ambiguity can be useful when it helps investigators challenge assumptions and consider alternative explanations.
- AI could help investigators understand where conclusions are supported by evidence and where uncertainty remains, while human judgement remains central to the investigative process.
Watch highlights from the episode
Is AI Breaking the Rules?
Should We Be Fearful Of AI?
