AI is reshaping software engineering, but the changes are more complex than a simple narrative of disruption or enablement. Drawing on interviews with leading CTOs and data from LinkedIn and the Tech Council of Australia, the report presents a nuanced view of AI’s current and future impact on engineering talent, culture, and capability.

Key Themes:

1. The Current Landscape

• Australia's AI workforce grew from 800 in 2014 to over 33,000 in 2023.
• Despite concerns, AI has not yet reduced engineering jobs—most titles still show positive growth.
• Roles specific to AI (e.g., Generative AI Engineers) are increasing rapidly, while more generalist roles (e.g., Full Stack Engineers) show stagnation or decline.

2. CTO Sentiment: Cautious Optimism

• Most CTOs agree that AI is a tool, not a replacement—currently helpful for scaffolding code, but not yet reliable in production.
• AI's speed of advancement is underestimated, posing a risk of obsolescence for companies and individuals who fail to adapt.

3. The ‘70% Problem’

• AI often delivers ~70% accurate code, but struggles with the remaining 30%—contextual judgement, debugging, and domain-specific knowledge remain human strengths.
• There’s concern that junior engineers may over-rely on AI and fail to build foundational skills.

4. The Knowledge Challenge

• As AI increasingly abstracts away complexity, domain-specific knowledge is at risk of erosion.
• Historical economic theory (e.g., Hayek’s Knowledge Problem) is referenced to show that AI, like central planning, may mismanage distributed knowledge if over-trusted.

5. Conformity & Cognitive Offloading

• Engineers may blindly follow AI-generated solutions, especially junior staff, leading to loss of critical thinking and creativity.
• Experiments like the Asch Conformity Test are referenced to warn against groupthink in engineering decisions guided by AI.

6. Recruitment & Training Implications

• CTOs suggest hiring for creative, commercially-minded engineers who understand the business—not just technical skills.
• AI enables broader participation (e.g., self-taught “normies”), but companies must still invest in foundational training to maintain quality and innovation.

7. Risks of AI-Induced Stagnation

• Recursive model training could lead to “model collapse”, where generative systems regress by learning only from prior AI outputs.
• If not managed properly, AI may encode biases, limit heterodox thinking, and undermine innovation.

Recommendations for CTOs & Tech Leaders:

• Do not abandon training juniors. Basic software skills remain critical, especially for debugging AI-generated outputs.
• Use AI as augmentation, not substitution. Treat AI as a ‘copilot’—valuable, but still needing human oversight.
• Hire for creativity and business awareness. The most valuable engineers will be those who combine domain expertise with strategic thinking.
• Develop AI policies. Most teams lack clarity on responsible AI usage—urgent governance is needed.
• Prepare for long-term effects. Avoid “presentism”—think 10–20 years ahead to anticipate structural shifts in team capability and knowledge distribution.

AI is transforming software engineering, but its adoption must be strategic and human-centric. The next era of technology will belong to organisations that embrace AI while continuing to nurture talent, preserve institutional knowledge, and think independently.

‍