AI Engineer Melbourne 2026 Keynote Livestream | Day 2

Key Concepts

• Self-Determination Theory (SDT): A psychological framework identifying autonomy, mastery, and relatedness as the core drivers of human flourishing and motivation.
• Eudaimonia vs. Hedonia: The distinction between "flourishing" through actualizing one's capacities (Eudaimonia) versus seeking frictionless, easy pleasures (Hedonia).
• Dark Flow: A state of superficial engagement that mimics "flow" but leads to addiction and stagnation rather than growth (e.g., gambling or mindless AI interaction).
• Vibe Coding: A term describing the practice of interacting with AI to generate code without deep understanding or rigorous engineering, often leading to a false sense of productivity.
• Supervisory Engineering: A new paradigm where the developer’s role shifts from direct creation to directing AI agents and verifying their outputs.
• Self-Efficacy: The belief in one's own ability to accomplish tasks; identified as the strongest predictor of productivity and developer experience in the AI era.

---

1. The Psychology of Work and AI

Jeremy Howard emphasizes that the current AI revolution is not just a technical shift but a psychological one. Drawing on 50 years of research, he argues that humans are naturally agentic and motivated by growth. However, modern AI tools can either support this growth or diminish it.

• The "Decay" Risk: When AI is used to outsource thinking, it creates an "illusion of control." This leads to "vibe coding," where developers feel productive because they are generating output, but they lose the ability to verify or understand the underlying systems.
• The "Growth" Opportunity: AI can be used to augment human intellect—a concept dating back to Douglas Engelbart’s "Mother of All Demos" (1968). By using AI to explore, debug, and experiment, developers can achieve a deeper mastery of their craft.

2. Research Findings: The Lived Experience of Engineers

Howard presented findings from a longitudinal study of software engineers across 28 countries:

• Shift in Tasks: There is a statistically significant shift from "creation-focused" tasks to "verification-focused" tasks.
• The Productivity Paradox: While 84% of engineers report higher productivity, 27% report a decline in "developer experience" (cognitive load, flow state, and feedback loops).
• Self-Efficacy as a Predictor: The most critical factor for success is self-efficacy. Engineers who believe in their ability to learn and adapt are 10 times more likely to report productivity gains.

3. Methodologies for Augmenting Intelligence

Howard advocates for using AI as a tool for "effortful craft" rather than a replacement for it.

• The "Solvit" Approach: Howard demonstrated his tool, Solvit, which allows users to load complex papers or codebases and engage in a dialogue to understand them deeply.
• Step-by-Step Learning: Instead of asking AI to "do it for me," the methodology involves:
  1. Deconstruction: Asking the AI to explain specific figures or concepts.
  2. Hypothesis Testing: Proposing a solution to the AI and having it verify or debug the logic.
  3. Re-implementation: Manually rebuilding components (e.g., CSS styles or recursive language models) to ensure internal understanding.

4. Future Archetypes for Developers

Howard outlines three potential futures for software engineers:

1. The Artisanal Developer: High-skill, focused on safety-critical or regulated environments.
2. The Clerical Coder: A role to be avoided; involves uncritically accepting AI-generated code, leading to low job satisfaction and skill atrophy.
3. The Orchestrator (Code Conductor): The ideal path, where engineers build software at scale by directing agents while maintaining deep domain expertise.

5. Notable Quotes

• "The human spirit can be diminished or crushed, and that individuals sometimes reject growth and responsibility." — Quoting the Self-Determination Theory review paper.
• "The people getting you to use AI don't care about your autonomy and mastery. They care about your outputs." — Jeremy Howard, warning against corporate-driven "token-maxing" metrics.
• "We've amplified the intelligence of a human by organizing his intellectual capabilities into higher levels of synergistic structuring." — Douglas Engelbart (1962), cited as the foundational mission for modern AI engineering.

6. Synthesis and Conclusion

The core takeaway is that AI is neither inherently good nor bad for the psyche; its impact depends on the user's intent. To avoid "AI burnout" and the trap of "dark flow," engineers must intentionally design their workflows to prioritize mastery and autonomy. By treating AI as a partner in an "effortful" learning process rather than a shortcut, developers can navigate the current transition and maintain their professional pride and joy. The future of engineering is not about being replaced, but about becoming a "code conductor" who designs the systems that build the future.