As LLMs get better and better equipped with good tooling, most AI evangelists are selling the “AI will replace our workforce” narrative. The most notorious is probably Andrej Karpathy who came up with the idea of ‘vibing’, and quite often posts “Look how much my AI did without me touching anything”. Such things, in my opinion, are purely marketing gimmicks. I’d say, typically a human + AI workflow is surely better and often faster than running Claude Code and watching it struggle for 2 hours. Other similar narratives go like “English is going to be the only programming language” and “no one is going to read/write code anymore”. Another one: “Everyone is going to be a manager, no more engineers.”
However, from my experience, and of several folks I know personally, AI has been only a great amplifier and augmenter of skills and talents that folks already have. And of incompetencies too. Somebody recently described a project to me and asked “Can Claude Code do this?”, and I replied: “Depends on who uses Claude Code”. One can use Claude to generate 10000 lines of React slop or 100 lines of vanilla JS that does the same thing. It depends.
Hot take: AI only makes things faster if you know what you’re doing
One of the ways I’ve been describing this experience of mine to other folks is the Iron Man narrative. Why is the Man still inside in the Iron when he has JARVIS to control the Iron and a whole fleet of Irons? I asked Gemini this question, and what we can learn about AI replacing people from this analogy. I’m paraphrasing the conversation below.
Me:
Why does Tony Stark stay inside the suit in Iron Man and all the other movies/comics when he can just command it remotely?
Gemini:
While Tony Stark does frequently use remote-operated suits (like the “Iron Legion” in Iron Man 3 or the “Sentry Mode” in Spider-Man: Homecoming), there are several strategic and narrative reasons why he usually chooses to stay inside the armor.
1. Latency and Signal Interference
In a high-stakes dogfight or a battle against superpowered beings, even a millisecond of lag can be fatal. Remote piloting relies on a data link that can be jammed, hacked, or physically obstructed by deep-underground bunkers or atmospheric interference. By being physically present, Tony ensures a zero-latency connection between his brain and the suit’s HUD and movement systems.
2. The “Human Element” and Intuition
Tony often argues that the suit is an extension of himself, not just a drone. Artificial intelligence (like JARVIS or FRIDAY) is excellent at calculations, but it lacks the unpredictable human intuition needed to outsmart an opponent. Being in the suit allows him to make split-second “gut” decisions that an algorithm might not prioritize, such as self-sacrifice or creative problem-solving.
3. Survivability and Life Support
The Iron Man suit isn’t just a weapon; it is a high-tech life-support system. In the early films, it literally kept his heart beating. Even after his surgery, the suit provides:
Physical Protection: Tony is a “glass cannon”—a genius human who is physically vulnerable. The safest place for him on a battlefield is inside the most durable object in the room.
Environmental Control: It allows him to survive in the vacuum of space, underwater, or in toxic environments where he might need to lead or command from the front.
4. Accountability and Leadership
From a character standpoint, Tony struggles with the idea of “automated” protection (which is the central conflict of Avengers: Age of Ultron). He feels a moral obligation to be the one pulling the trigger or taking the hits. Leading from a remote bunker would distance him from the consequences of his actions—a detachment he spent his entire post-Afghanistan life trying to avoid.
Me:
Apply this analogy to AI models and tell me how they will impact work
Gemini:
The “Pilot vs. Remote” analogy perfectly maps onto the current evolution of AI in the workplace. Just as Tony Stark shifted from being the sole operator to managing an “Iron Legion,” we are moving from AI as a tool to AI as an agent.
Here is how that analogy translates to the future of work:
1. The “In-the-Suit” Era: AI as an Exoskeleton
In this phase, the human is the pilot. You aren’t “remote”; you are using AI to augment your own physical or cognitive limits.
The Impact: Increased individual “torque.” A single developer, designer, or auditor can do the work of a small team because the “suit” (AI) handles the heavy lifting, data processing, and rote movements.
The Key: You still bear the risk and provide the intuition. If the “suit” fails, you are the one on the ground.
2. The “Iron Legion” Era: AI as Autonomous Agents
This is the “remote command” stage. Instead of doing the work, you are the Commander. You deploy “suits” (specialized AI agents) to handle specific tasks—one for research, one for drafting, one for testing.
The Impact: Shift from Execution to Orchestration. Your value isn’t in how well you “fly” the suit, but in how well you coordinate the fleet.
The Risk: Just like Tony’s remote suits, this introduces “latency” in quality control and the risk of “signal interference” (hallucinations or misaligned goals).
3. The “Human Element” (The Moral Center)
The biggest impact on work will be the re-valuation of intuition. As AI becomes better at “remote” execution (the calculations), the “pilot in the suit” becomes more important for the things AI can’t simulate:
Accountability: AI can generate a solution, but a human must own the consequences.
Edge Cases: In a “battle” (a complex, unprecedented business or technical crisis), the algorithm might lack the “gut feeling” to pivot.
Empathy: Human-to-human leadership remains the one thing you can’t command remotely through a drone.
The “Ultron” Warning: Tony Stark’s biggest mistake was trying to “put a suit of armor around the world” without enough human oversight. In a professional context, over-reliance on automated “remote” AI can lead to Technical Debt and a loss of specialized knowledge. If no one knows how to fly the suit manually, you’re in trouble when the signal goes down.
P.S.: Tony Stark risks his life to be the man in the suit despite having the most advanced AI on the planet. Hands-on engineering is often zero-risk and high-return (higher return when AI is skillfully used).