Robert Koch

McDonald’s new ad implies we are going to run out of chicken due to exhausting natural reserves.

artxscience 2025 #25 - reNEW 71

Check out my friend's stunning entry in the @reNEW_Global #ARTxSCIENCE Competition! Dive in and cast your vote!

renew.science/artxscience...

The cure is a real AI strategy.

The question isn’t “How do we stop Shadow AI?”
The question is “How do we harness it?”

The best approach:
1. Identify where it’s happening
2. Observe how teams are using it
3. Embrace and extend the practices that drive real results

Shadow AI is a symptom.

If your company hasn’t provided a sanctioned AI option, your employees will find one.

The path forward isn’t punishment it’s enablement.
Document it. Embrace it. Extend it.

Because if your highest-performing team is secretly using AI, that’s not a problem.

That’s a signal. 🚨

Let’s be honest: employees don’t turn to Shadow AI because they’re reckless. They do it because:
🏃 It makes their jobs faster
🏋️‍♀️ It makes hard tasks easier
🚚 It helps them deliver more with less effort

I’m curious: what’s the bigger challenge in the energy transition - technical innovation, policy frameworks, or the sheer number of skilled engineers to make it happen?

#NetZero #FutureOfWork #Engineering

Combine those two numbers, and the question becomes:
- Do we have the engineering capacity to deliver the transition we need?
- Or will the bottleneck not be technology, but talent?

⚡ Quick stat that hit me today: Australia graduates ~17k engineers per year. China graduates ~1.3 million.

At the same time, ~70% of the infrastructure that will exist in 2050 isn’t built yet.

If you’re a CISO or CTO, the question isn’t “Do we allow AI?” it’s “Do we know where and how our people are already using it?”

The result? Confidential data exposed to vendors with zero SLAs or compliance clearance.

Shadow AI doesn’t just accelerate work it can accelerate data leaks.

Shadow AI isn’t just about productivity. It’s about security.

I’ve seen companies discover, after the fact, that employees were:
👉 Using external LLMs in other countries to summarise sensitive documentation
👉 Running proprietary code through unapproved AI tools

Because ignoring Shadow AI doesn’t make it disappear it just makes the risks invisible.

This isn’t inherently a bad thing — but if it’s undocumented, it introduces security, compliance, and continuity risks.

Leaders: Don’t just ask if AI is being used. Ask how it’s being used.

How do you know if Shadow AI is creeping into your business?
Here’s a simple litmus test:

An employee who previously struggled with a task suddenly delivers flawless results with surprising speed.

AI might be doing the heavy lifting.

👉 2026 won’t just be about AI models. It will be the year of MCP and interaction protocols, where #GenAI becomes not just a thinking partner but a doing partner.

👉 Simplification of complexity: Instructions like “Go into this hazardous zone and take measurements” will no longer require deep technical expertise. With MCP, even an entry-level employee could safely command robots to execute these tasks.

👉 Real-world integration: Agents won’t just be digital assistants—they’ll control physical systems. Cars, robotic arms, even ships could be safely directed using simple natural language.

👉 Cross-industry adoption: Where GenAI started in software, MCP-style servers will spread into finance, healthcare, and even food services.

In 2025, we’ve seen agents dominate the conversation in AI/ML. But by 2026, the real shift will come from how these agents connect with real-world workloads through standards like the Model Context Protocol (MCP).

Here’s what I expect:

🚀 The future of generative intelligence isn’t just smarter models - it’s agents that can understand and act.

The risk: Leaders may have no idea how deeply it’s already embedded in their workflows.

Question for leaders:
Do you think your teams are fully transparent about their AI use or is Shadow AI already part of your daily operations?

Why? Because once employees figure out how AI can make their jobs faster and easier, they won’t wait for company policies to catch up. They’ll just use it.

The reality: Shadow AI is becoming the default for many workers.

Shadow AI isn’t just happening — it’s everywhere.
Deloitte recently found that 27% of employees are using AI without their manager knowing. From my perspective, that number is probably much higher.

Never, Ever Make Recursive Lambda Calls | Kochie Engineering I’m not sure why this still needs repeating, but please—stop writing recursive Lambdas.

If your AI copilot suggests a recursive Lambda, throw your laptop out the window (or at least hit undo).

blog.kochie.io/articles/13...

📉 State management in your Lambda? Also a trap. Use Step Functions or SQS if you want retries, loops, or sanity.

I once saw a recursive Lambda spiral into a 1000 invocation death loop. The CloudWatch bill alone could’ve funded lunch for a large team… for a month.

Every nested invocation compounds your bill, slows down performance, and amplifies failure risk.

💸 AWS Lambda charges per invocation and duration. That means recursive Lambdas equal exponential spend for linear work.

New blog post. I really wanted to rant about some silly things I've seen people do with Lambda functions.

It’s 2025 and somehow this still needs to be said.

Recursion is elegant in theory BUT, in the cloud, it’s a costly disaster waiting to happen.

There’s still time to grab your ticket – whether you come for a day or the full experience! Will I be seeing you at #CSE25?

Book your seat: Climate Smart Engineering | Engineers Australia
www.engineersaustralia.org.au/learning-an...

CSE25 is just around the corner!
I’m excited to be chairing the Technical Program for Climate Smart Engineering 2025, happening in Adelaide 27–28 August. With over 100 speakers across two days, the program dives into decarbonisation, innovation, climate resilience and more.