Tony Fadell on Building Taste in the AI Era: Why Product Judgment Still Trumps Algorithms

TL;DR

• Taste is a learnable skill, not innate talent: Tony Fadell argues that great product judgment comes from deliberate practice—experiencing products deeply, understanding what works and why, and building an internal reference library of excellence.
• AI amplifies execution but can't replace human judgment: While AI accelerates prototyping and iteration, the critical decisions about what to build, for whom, and why still require human taste, empathy, and strategic thinking that no algorithm can replicate.
• The "taste stack" matters more in the AI era: As AI commoditizes technical execution, the differentiator becomes your ability to make nuanced judgment calls about user experience, emotional resonance, and cultural fit—skills that require years of intentional cultivation.
• Product builders must actively resist AI-driven mediocrity: Without strong taste and judgment, teams risk letting AI tools guide them toward statistically average solutions rather than breakthrough products that genuinely matter to users.

The Uncomfortable Truth About AI and Product Development

We're living through a peculiar moment in product development. AI tools promise to democratize creation, compress timelines, and eliminate technical barriers. Yet the products that truly break through—that change behavior, that people love—aren't getting easier to build. If anything, they're getting harder.

Tony Fadell, the designer behind the iPod and iPhone, recently sat down with Lenny Rachitsky to discuss something most product builders are quietly worried about: whether taste and judgment still matter when AI can generate infinite variations of anything. His answer should make every product manager, designer, and founder pay attention.

The short version? Taste matters more than ever. But not for the reasons you might think.

Why Taste Isn't What You Think It Is

Most people treat taste as mystical—something you either have or don't. Fadell demolishes this notion immediately. Taste, he argues, is a skill you build through deliberate exposure and critical analysis. It's not about having "good taste" in some abstract sense. It's about developing a sophisticated internal library of what works, what doesn't, and most importantly, why.

Think of it like training a model, except the model is your brain. You feed it thousands of examples: products you use, experiences you have, designs you study. But unlike passive consumption, you're actively interrogating each example. Why does this interface feel intuitive? What makes this physical product satisfying to hold? Why did this feature fail despite being technically impressive?

Fadell describes his own process: constantly using products, even ones outside his domain, and asking hard questions about every design decision. Not just "do I like this?" but "what problem is this solving, and is this the right solution?" This kind of active analysis builds what he calls a "taste stack"—layers of judgment that inform every decision you make as a builder.

For product managers working with AI, this reframing is critical. You're not trying to develop some ineffable artistic sensibility. You're building a decision-making framework based on deep pattern recognition across thousands of product experiences.

The AI Execution Paradox

Here's where it gets interesting for those of us building AI products. Fadell acknowledges that AI is transforming the execution layer of product development. Need to generate design variations? AI can do it. Want to prototype faster? AI accelerates it. Need to analyze user feedback at scale? AI handles it.

But—and this is the critical insight—AI's strength in execution actually increases the importance of human judgment at the strategic layer.

When you can generate a hundred design variations in an hour, the bottleneck isn't creation anymore. It's knowing which variation to pursue. When you can build features faster than ever, the constraint becomes understanding which features actually matter. When you can analyze mountains of data, the challenge is knowing what questions to ask.

This is where taste becomes your competitive advantage. AI tools are becoming commoditized rapidly. Every team has access to similar capabilities. What distinguishes great products from mediocre ones isn't the sophistication of the AI under the hood—it's the judgment calls made by the humans directing it.

My Take: We're Training AI to Optimize for the Wrong Things

I think we're sleepwalking into a trap, and Fadell's insights crystallize why I've been uneasy about how many teams are integrating AI into their product development process.

The problem is this: AI tools, by their nature, optimize for patterns in existing data. They're incredibly good at identifying what's statistically common, what's worked before, what fits established patterns. But breakthrough products—the iPod, the iPhone, the products that actually matter—don't come from optimizing existing patterns. They come from having the judgment to break them in exactly the right way.

When you let AI guide too much of your product development without strong human taste as a filter, you risk building products that are technically competent but fundamentally uninspired. You get the statistical average of what exists, refined and polished, but not transformed.

I've seen this firsthand with AI product teams. They move faster, ship more features, iterate more quickly. But when you actually use the products, something's missing. They feel like they were designed by committee—or worse, by algorithm. They lack the opinionated point of view that makes products compelling.

The solution isn't to reject AI tools. It's to recognize them for what they are: powerful execution engines that need strong human judgment to direct them. You need taste to know when to override the AI's suggestions, when to pursue the statistically unlikely option, when to make the bold call that no algorithm would recommend.

Building Your Taste Stack in the AI Era

So how do you actually develop this judgment? Fadell's approach is surprisingly systematic for something often treated as intuitive:

1. Use Everything Intentionally

Don't just use products—study them. Fadell talks about deliberately using products outside his comfort zone, paying attention to every interaction, every decision the designers made. For AI product builders, this means using AI products critically. Why did ChatGPT's interface work when so many chatbots failed? What makes Claude's interaction model feel different? What's working in Midjourney that isn't in other image generators?

The key is active analysis. After using a product, write down what worked and what didn't. Not just "I liked it" but specifically why. What problem did it solve elegantly? Where did it create friction? What would you do differently?

2. Build Your Reference Library

Fadell describes maintaining a mental catalog of excellent execution across domains. A physical product with perfect tactile feedback. A digital interface with flawless information hierarchy. A service experience that anticipated needs beautifully. These references become touchstones when you're making your own decisions.

For AI products, this means studying not just AI interfaces but any product that solved a complex problem elegantly. How did Google Maps make navigation intuitive? How did Stripe make payments developer-friendly? These patterns transfer.

3. Understand the "Why" Behind Decisions

This is where most people stop too early. It's not enough to notice that something works. You need to understand the reasoning behind it. Why did Apple put the clickwheel on the iPod instead of buttons? Why does the iPhone's home button placement matter?

When you understand the strategic reasoning behind design decisions, you can apply similar thinking to new problems. You're not copying solutions—you're learning decision-making frameworks.

4. Practice Making Judgment Calls

Taste develops through exercise. Fadell emphasizes that you need to make decisions, see the results, and learn from them. This is hard with AI products because the feedback loops are often longer and noisier. But it's essential.

Start small. Make judgment calls on feature prioritization. Decide which AI capability to expose to users and which to keep hidden. Choose how much control to give users versus how much to automate. Then watch what happens. Did users respond as you expected? Why or why not?

The Creativity Challenge: Humans + AI

Fadell raises a crucial point about creativity in the AI era: the risk isn't that AI will replace human creativity, but that humans will stop exercising their creative muscles because AI makes it easier to accept "good enough."

This is already happening. I see product teams using AI-generated designs without pushing them further. Using AI-written copy without injecting personality. Using AI-suggested features without questioning whether they're actually solving the right problem.

The antidote is intentionality. Use AI as a starting point, not an ending point. Let it handle the mechanical work—generating variations, exploring possibilities, executing on decisions. But reserve the creative judgment for humans. Push past the first AI-generated option. Question the assumptions embedded in AI suggestions. Use your taste to guide the AI toward something genuinely novel.

Fadell's work on the iPod and iPhone happened in an era of intense constraint. Limited technology, limited screen space, limited battery life. Those constraints forced creative breakthroughs. In the AI era, we have the opposite problem: too many possibilities, too few constraints. Your taste becomes the constraint that focuses creativity toward meaningful innovation.

What This Means for Product Builders Today

If you're building AI products—or any products in the AI era—here's what Fadell's insights mean practically:

First, invest in your own taste development. This isn't optional professional development. It's core to your role. Spend time using products critically. Study design decisions. Build your mental library of excellence. This is as important as learning new AI tools or frameworks.

Second, resist the pull toward AI-driven mediocrity. When AI suggests the statistically likely solution, ask whether that's actually the right solution. Be willing to override the algorithm when your judgment says something different. The best products often come from decisions that look wrong in the data but right in human experience.

Third, use AI to amplify your judgment, not replace it. Let AI handle execution speed and variation generation. But keep the strategic decisions—what to build, for whom, why—firmly in human hands guided by human taste.

Fourth, build taste-development into your team culture. Make it normal to discuss why products work or don't work. Create space for team members to study and analyze products together. Share your reference libraries. Make taste development a team sport.

Finally, remember that taste compounds. The judgment you build today informs decisions you'll make years from now. Fadell's ability to make brilliant calls on the iPhone came from years of building taste through earlier products. Start now. The taste you build in this AI era will be your competitive advantage for the next decade.

The Long Game

Fadell's career arc illustrates something crucial: the products that matter most take years to develop, both the products themselves and the taste required to create them. The iPod wasn't his first product. The iPhone built on everything that came before.

In an era where AI promises to compress timelines and democratize creation, it's tempting to think the long game doesn't matter anymore. But Fadell's insights suggest the opposite. The long game matters more than ever because taste—the kind that creates breakthrough products—only develops over time.

AI can make you faster. It can make you more productive. It can help you execute more efficiently. But it can't give you the judgment to know what's worth building in the first place. That still requires human taste, human creativity, and human judgment developed through years of intentional practice.

For product builders in the AI era, this is both challenging and liberating. Challenging because you can't shortcut taste development. Liberating because it means your human judgment remains your most valuable asset—the thing that can't be automated, can't be commoditized, can't be replicated by the next AI model.

The question isn't whether taste matters in the AI era. It's whether you're investing in developing yours while everyone else is distracted by the shiny new tools. Because the builders who combine strong taste with AI capabilities won't just build better products. They'll build the products that actually matter—the ones people love, the ones that change behavior, the ones we'll still talk about decades from now.

That's the real lesson from the father of the iPod and iPhone. Technology changes. Tools evolve. But the human judgment required to create products that genuinely matter? That's timeless. And in the AI era, it's more valuable than ever.