Has this happened to you recently? A coworker shares a polished design document, detailed technical strategy, or a beefy proof-of-concept. The presentation is clean, the logic is sound, and the prose is eloquent. But when a simple question probes their reasoning—"What's the value of using an in-memory cache here?"—the conversation stalls. The reply is vague, evasive. It lacks conviction.
I asked my work friend about this in a recent one-on-one. "It's happening all the time" they said. We are observing a dangerous trend taking root: the quiet acceptance that conviction is no longer necessary. An assumption that smart-sounding arguments are sufficient, and you no longer need to believe what you present, understand its origins, or build conviction with your peers. This is the rise of empty eloquence, and it threatens the joy and future of building delightful products in teams.
On Conviction
Conviction isn't a flash of insight. It's a slow burn, the byproduct of friction.
Conviction starts with a hypothesis—an approach, a design decision, a belief about your user. You share it with someone you trust for feedback. You hear their challenges. When your arguments don't hold, you learn. Or you will if your strong opinions are held with humility. You don’t expect to end up where you started.
Earning the respect of your peers, learning from their challenges, and building something of value together is intoxicating. Conviction thrives on constructive conflict. This kind of conflict is the prerequisite for a healthy team. Constructive conflict is the primary mechanism for forging the trust and mutual respect that high-performing teams are built on.
These are also desirable difficulties for individuals—effortful struggle that cements mastery, not shortcuts it. It's a durable understanding forged in the discomfort of defending our reasoning to skeptical—and respected—colleagues.
When we do this, something valuable is happening beyond the outputs: we're building relationships. Strong relationships are the primary predictor of joy and satisfaction. AI offers eloquent arguments without demanding conviction. If we take this shortcut, conviction isn’t the only thing we’ll lose.
The Problem of Empty Eloquence
Theologian John Piper captured what’s missing when he reflected on machines that can replicate human speech:
"These machines honor me with their lips and their bytes, but their heart is far from me. Computers do words better than you - but they don't feel anything... the universe is created to have people who [feel] the worth, glory, beauty, and wonder of grace."
You don't have to share Piper's theological framework to recognize the profound truth he's pointing toward. AI excels at producing words that sound authoritative, reasoned, even emotionally moving. But eloquence cannot replace conviction. Conviction is the residue of a struggle—the intellectual and emotional work of wrestling with an idea until it can be trusted.
Passing off AI-generated work as your own is offering empty words devoid of the belief, doubt, and discovery that makes collaboration meaningful. The shortcut bypasses the work that builds trust.
Our colleagues aren't processing units. They are partners in the deep work of building something meaningful. When we offer them empty eloquence, we're undermining the joy that makes our work valuable and sustainable. This erosion has real costs.
What We Lose
Impotent meetings and shallow technical discussions aren’t the only risks. When we shortcut conviction, we lose things fundamental to strong teams, delightful products, and our personal growth.
Retention. You don't remember arguments with computers. You remember the debate at the whiteboard when Sarah challenged your assumptions. The relational component of that struggle is what cements learning in a way solo interaction with an LLM cannot.
Trust. Professional relationships are built on intellectual honesty and demonstrated thinking. When people know you as someone who thinks deeply and can explain your ideas, you build compounding professional capital. Eloquent words without conviction erode trust faster than you can rebuild it.
Joy. Not caring about your ideas, not building conviction through community struggle and debate, makes work less satisfying. Collaborative idea development makes challenging work sustainable, and enjoyable, long-term.
Serendipity. Colleagues constantly acquire new context and share insights, connections, and opportunities. They remember your interests, your expertise, and your past arguments. Humans still have a massively larger context window and capacity for retrieval than LLMs. Your colleagues will proactively reach out to you with things of value—but only if you've built relationships through meaningful collaboration.
Augment or Abdicate?
I want to be clear: I'm not arguing against using AI. I do so extensively, and I think most engineers should lean in more, not less.
My case is to use AI with intention. The hype cycle promises infinite productivity, but the data and our own experience point to a sober reality: there’s an automation plateau. Early gains are significant, but they flatten out as hidden costs—oversight, rework, and the erosion of team cohesion—begin to mount.
The asset that compounds is not your ability to automate, it is the relational capital you’ve invested.
How do we decide when to automate and when to abdicate to AI? Ask yourself: do I need efficiency, or do I need conviction? Using AI to augment our process is smart. Abdicating our responsibility to build conviction, forge relationships, and own our judgments is a shortcut we cannot afford.
Journalist Ezra Klein touches on why this is so critical, contrasting deep reading with AI summaries:
Part of what is happening when you spend seven hours reading a book is you spend seven hours with your mind on this topic. The idea that [an AI] can summarize it for you... is that you didn't have the engagement. It doesn't impress itself upon you. It doesn't change you. What knowledge is supposed to do is change you, and it changes you because you make connections to it.
— Ezra Klein on How I Write
AI excels at gathering context, surfacing counterarguments you might have missed, and helping you organize complex information. What it can't do is care about the outcome, build relationships, or develop conviction.
We're trying to reduce toil, if we find that it reduces actual context [understanding] we will revert.
— Farhan Thawar, Head of Engineering, Shopify
I regularly use AI to keep documents up-to-date. I supply transcripts of calls and have it compare what was discussed with what's currently written. The AI identifies decisions that have drifted, suggests updates, and writes fantastic meeting notes with key decisions and learnings up top. This frees me to focus on strategic thinking and relationship building.
This isn't a perfect science. It's a continuous, deliberate practice of asking: "What is the real work to be done here?" Sometimes the real work is just getting the code written. Often, the real work is the human struggle that happens along the way. That’s the part we can’t afford to automate.
The Path Forward
The teams that thrive won't reject AI on principle, nor will they automate every task. They'll use AI to handle toil so they can spend more time on the hard problems that require human judgment and interaction—collaboration, learning, and the intrinsic value in being challenged by another person.
Be radically transparent. When I collaborate with AI, I say so: "I had Claude help me compile some of this perspective, let me know what you think." This doesn't weaken your position; it builds trust with humility, honesty, and an invitation to collaborate.
My thinking is evolving, but my conviction on the core risk is firm: if we outsource the struggle, we outsource the trust, the joy, and the relationships that make the work worth doing.
I'm having more fun writing software and doing knowledge work with AI assistance than I have in years. You don’t have to be an AI Optimist or an AI Doomer. We can embrace these tools thoughtfully while protecting what makes us human—our capacity for conviction, connection, and to feel the worth, glory, beauty and wonder of grace.
When we get the balance right—when we use AI for augmentation and preserve space for the desirable difficulties that build conviction—we create something beyond better software. We preserve the joy and connection that comes from intellectual struggle shared with people we respect.