Cloud Dialogues

When AI Leaves the Data Centre: Robots, Power Grids & the New Attack Surface 🤖⚡

AI is moving rapidly beyond chat windows and into the physical world — powering robots, vehicles, industrial systems, edge devices and critical infrastructure.

In this episode of Cloud Dialogues, Georgia and Matt are joined by Alex Turner, founder of Admiral, to explore what happens when increasingly capable AI systems meet the messy reality of hardware, legacy networks and infrastructure that cannot simply be rebooted.

They discuss:

* 🧠 Why frontier AI labs are talking about slowing development — and whether there are commercial incentives behind the message
* 💣 The possibility of deploying “time bombs” into code that may be readable, but unnoticed
* 👀 Why telling an AI system what not to do is not the same as building a real guardrail
* ⚡ How AI could lower the expertise required to attack power grids and critical infrastructure
* 🏭 Why legacy industrial systems and SCADA networks may be particularly exposed
* 📡 How cheaper, more powerful edge hardware is bringing real-time AI into the physical world
* 🚛 What happens when a software update bricks an entire fleet of devices
* 🔥 Why autonomous devices need observability, controlled rollouts and a genuine kill switch
* 🤖 The future of robots, automated restaurants and AI-powered machines operating around us
* ⚙️ Whether smaller, type-safe decision models like JEV could become more useful than giant language models for certain operational tasks

The future of AI may not look like a chatbot.

It may look like a robot, a delivery vehicle, a power-grid controller or a machine quietly making decisions at the edge — often on someone else’s network.

And the uncomfortable question is: who is watching what happens when those systems go wrong?

You can find Alex on LinkedIn: https://www.linkedin.com/in/alexanderpaulturner and Admiral at https://admrl.co

What happens to software engineering when developers stop writing code and start  directing, evaluating and correcting AI?

In this episode, Georgia and Matt are joined by Annie Vella, former Distinguished Engineer at Westpac New Zealand, to explore how AI coding tools are reshaping software engineering.

Drawing on her research into AI-assisted development, Annie introduces the idea of “supervisory engineering”: the growing work of directing AI, evaluating its output, correcting mistakes and integrating generated code into complex systems.

They explore:

• The hidden cognitive load behind AI-generated code
• Why code review is becoming a major bottleneck
• What accountability means when engineers approve code they didn’t write
• Whether we’re automating the craft and enjoyment out of software development
• Why foundational engineering skills matter more than ever
• What the next generation of developers should be learning
• How small teams can capitalize on AI without losing essential human capabilities

It’s a conversation about the future of software development, but also about something much bigger: what happens to human judgment, learning and pride when more of our thinking is delegated to machines?

In Episode 42 of Cloud Dialogues, Georgia and Matt are joined by AI ethicist and The Mathpath founder Aubrey Blanche for a candid, provocative and surprisingly funny conversation about responsible AI in the real world.

We explore:

🔐 AI security and sovereignty
Frontier-model hype, open-weight models, enterprise security, data residency and why Microsoft doesn’t necessarily need the “best” product to win.

🧠 “More capable,” not “smarter”
Why anthropomorphising AI distorts how we understand - and interact with - the technology.

⚖️ What algorithmic discrimination actually looks like
Why “the evil computer did it” is rarely an accurate or useful diagnosis - and why understanding the mechanism matters if we genuinely want to fix it.

📈 The problem with productivity as the goal
Making broken processes faster may simply produce more garbage, more efficiently.

🎯 Incentives drive behaviour
Including the inevitable consequences of token-maxing dashboards, performance targets based on AI usage and telling employees they must use six tools whether they need them or not.

👩‍💻 Why eliminating junior roles is spectacularly shortsighted
Tokens may now be more expensive than interns - and without junior employees today, organisations won’t have senior experts tomorrow.

👥 Pair prompting and human expertise
How teams can use AI to accelerate delivery while preserving reasoning, shared knowledge, mentorship and technical skills.

💸 The economics behind the AI narrative
Why frontier-model companies need an enormous addressable market to justify their valuations - and why that makes “AI will replace all human labour” an extremely convenient story.

🌍 Who gets to shape AI?
Who decides what should be automated? Whose problems receive investment? Who benefits from the dominant narrative - and what possible futures disappear when we accept it without question?

A practical discussion about adopting AI intentionally, creating genuine business value and building systems that work without forgetting the humans inside them.

Also featuring:

✨ 100 beautiful AI-generated slides that may or may not be useful
🔥 The consultant you hire when you want the truth
🪙 Tokens being set on fire for the performance dashboard
🚗 AI’s ongoing refusal to wash Aubrey’s car

Matt and Georgia are joined by Georgie Healy, former Program Lead for Google’s AI Accelerator and creator of Attention Is All I Need, for a lively conversation about AI adoption, inclusion and the very human challenge of keeping up with rapidly evolving technology.

In this episode:

📰 The latest AI news
The group unpacks:

AI-related layoffs and whether AI is sometimes being used as a convenient excuse for restructuring

The growing security risks created by autonomous agents

New model releases and the impossible task of keeping up with them all

The increasingly complicated relationship between data centres, housing, energy and public infrastructure

🧠 Are we still using our brains?
AI can help us learn faster, explore new topics and stress-test our thinking - but it can also make it very easy to produce work we do not fully understand.

The group discusses:

Why critical thinking and causal reasoning matter more than ever

The risks facing graduates who rely on AI without understanding their own answers

Why lived experience and systems thinking remain incredibly valuable

How to use AI as a thought partner without creating yet more AI slop

🌍 Making AI feel more inclusive
Why do some people embrace AI immediately, while others feel overwhelmed, excluded or actively resistant?

Georgie argues that AI literacy should be pro-people, not simply pro-technology. Rather than forcing people to use AI for work, adoption may be more effective when people can use it to create something useful, meaningful or genuinely fun.

✨ Using AI to bring ideas to life
For the first time, people without traditional coding skills can build websites, create prototypes, explore hardware and turn ideas into something tangible - without necessarily needing a developer or CTO to get started.

👩‍💻 Who will shape the next era of technology?
Georgie explains why women building cyberdecks, Tamagotchis, e-readers and other wonderfully niche hardware projects have changed her view of who will lead the next wave of AI-enabled innovation.

🔥 Plus, several hot takes:

Georgia delivers an unexpectedly defence of M365 CoPilot

Matt assigns everyone an unreasonable amount of homework

Georgie explains why it is perfectly reasonable to feel complicated about AI

The group concludes that technology can be useful, concerning and fun - all at the same time

A thoughtful, practical and occasionally chaotic conversation about creativity, critical thinking, representation and what it really takes to bring people along on the AI journey.

(apologies - audio is not up to our usual standard, but this version is the best I could get it)
Guest: Lena Hall (Senior Director of Developer Experience at Akamai, formerly DevRel lead at Amazon Web Services and AI/Data Advocacy Director at Microsoft) joins Georgia and Matt to unpack the latest AI developments and what they mean for how we build software.

News Highlights

The episode kicks off with several major AI updates. The US government excluded Anthropic from a supplier list over concerns around WMD and surveillance policies - only for OpenAI to sign a government deal shortly after with similar language, raising questions about whether the decision was policy-driven or political.

Meanwhile, OpenAI released GPT-5.4, a reasoning-focused "thinking model" with tunable reasoning depth. Early feedback suggests stronger accuracy and less verbosity, though it consumes more tokens and is slightly more expensive to run.

The hosts also discuss a pledge from Meta, Microsoft, Google, and Amazon to fund new electricity generation to support AI infrastructure - a move framed as sustainability but widely seen as a practical response to AI's growing energy demand.

Finally, a new partnership between CVS Health and Google Cloud highlights a broader shift in hyperscaler strategy: AWS continuing to focus on horizontal infrastructure while Google invests more heavily in vertical AI solutions such as healthcare.

The Core Discussion: Engineering in the Age of AI

The main conversation explores how AI systems fundamentally challenge traditional software engineering practices.

Unlike deterministic systems, AI outputs exist on a spectrum of quality. A system may be operationally healthy yet still produce incorrect or harmful responses, creating a new category of production issues that are harder to detect and diagnose.

Lena argues that while organizations don't necessarily need entirely new AI platform teams, platform engineering must evolve. Teams need infrastructure for AI observability, evaluation frameworks, fallback mechanisms, and intervention controls. Without this foundation, individual product teams end up solving the same problems repeatedly.

A key takeaway is the need for clearer responsibility across three types of AI failure: capability issues owned by product teams, safety risks defined by leadership, and operational reliability managed by platform engineering.

The group also emphasizes the importance of product-level evaluation, focusing not just on model benchmarks but on whether AI actually works for real users. Effective evaluation frameworks measure capability, safety, and operational reliability, with scrutiny increasing for higher-risk applications.

For organizations adopting AI, Lena recommends a gradual approach: start with assistants for narrowly defined tasks, move to supervised agents, and only introduce autonomous systems once observability and governance are mature.

AI and the Human Factor

The discussion ends with the impact of AI on developers themselves. Engineers are spending less time writing code and more time making high-level decisions about architecture, system behavior, and trade-offs. While this can increase productivity, it also raises cognitive load and shifts responsibility toward more experienced engineers reviewing large volumes of AI-generated output.

Cool AI Pick

Lena's pick is Codex Spark, an ultra-fast model designed for executing well-defined tasks. Her preferred workflow combines reasoning models like GPT-5.4 for planning, then handing execution to Spark - highlighting a broader trend toward specialized models working together in AI development pipelines.