AI is a tool
“AI is a tool, just like other tools we use. And it’s clearly a useful one.”
Linus Torvalds’ recent pragmatic embrace of AI within the Linux kernel development ecosystem captures an essential truth: AI is neither magic nor an existential threat to the craft of technical work. It is an abstraction layer. Just as software engineering transitioned from raw machine code to assemblers, compilers, and IDE autocompletion, generative models and LLMs represent the next logical step in human-tool evolution.
I align strongly with this pragmatic stance. At work, AI tools function as high-velocity catalysts, accelerating data exploration, automating routine scaffolding, and surfacing non-obvious edge cases. Yet, treating AI as a tool rather than a replacement highlights two vital dynamics we must navigate: how knowledge work will adapt in the short-to-medium term, and the long-term societal implications for children growing up in an AI-saturated world.
Work in a Pre-AGI Era: Evolution, Not Extinction
There is a persistent narrative that AI will render entire professions obsolete overnight. However, short of achieving true Artificial General Intelligence (AGI, a machine capable of autonomous reasoning, context synthesis, and self-directed judgment across arbitrary domains), the reality of the workplace looks quite different.
Rather than wholesale replacement, we are observing a process of role evolution:
- Embedded Scaffolding: LLMs and specialized models are becoming deeply integrated into existing developer tooling, databases, and enterprise platforms. They handle repetitive setup, boil down dense documentation, and write unit tests.
- Elevation of Fundamentals: When code or analytical pipelines can be generated in seconds, the bottleneck shifts from syntactic execution to architectural integrity, verification, and domain logic.
- The Responsibility Constraint: As Linus pointed out, generating code or output is effortless, but you remain 100% accountable for what you ship. An LLM cannot be held responsible for system failures, inaccurate financial models, or security vulnerabilities.
In a pre-AGI landscape, workers who leverage these sophisticated tools will simply outperform those who resist them, while the core value of human expertise (critical thinking, domain understanding, and accountability) remains paramount.
The Real Worry: What About the Next Generation?
While adults with established skills can easily treat AI as a productivity multiplier, the implications for children and young adults learning the ropes are far more concerning.
The central paradox of AI-assisted learning is simple: To effectively evaluate AI outputs, you must first master the fundamentals manually.
- The Missing Struggle: Mastery comes from friction: debugging an elusive bug, working through raw mathematical derivations, or wrestling with structured problem-solving. If young learners delegate that friction to AI too early, they risk developing a superficial understanding of how complex systems actually function under the hood.
- The Entry-Level Dilemma: As AI automates junior-level tasks (such as writing basic code boilerplate or drafting initial summaries), the traditional path for entry-level workers becomes harder to navigate. How do we train the senior architects, lead engineers, and principal analysts of tomorrow if the “apprentice” tier of work is abstracted away?
- Cognitive Reliance: There is a genuine danger of cognitive atrophy if children are taught to rely on AI for synthesis and creativity before developing their own critical thinking faculties.
Moving Forward: Guarding the Craft
Navigating this transition successfully requires a balanced approach to adopting AI in our daily work and personal lives:
- Own the Output: Treat AI suggestions as raw drafts. Never commit, deploy, or sign off on work you cannot explain line-by-line or concept-by-concept.
- Teach the “Why”, Not Just the “How”: For the next generation, education must double down on core fundamentals, logic, and problem decomposition before introducing automated assistance.
- Focus on Domain Depth: As baseline technical skills become easier to automate, cross-disciplinary expertise, domain knowledge, and clear communication will become the primary differentiators.
AI is here to stay, and embracing it as a powerful, evolving tool is the most rational path forward. But as we integrate it into our workflows, we must ensure we don’t trade away the deep understanding and problem-solving resilience that future generations will need to build what comes next.