People Want AI to Help Them Grow, Not Just Get Things Done. Almost Nobody Is Solving for That.

Long before AI showed up in most people’s lives, it already had a reputation. For a lot of people, that reputation is Skynet, the AI from the Terminator movies that decides to eliminate humans by starting a nuclear war. Our fears today are a bit more subtle: not that AI will destroy us, but that it’ll quietly take over people’s jobs and replace their judgment, one convenient answer at a time.

Anthropic recently published one of the largest studies of its kind — 81,000 people interviewed, conversationally, about what they actually want from AI. Here’s the finding that should matter to every brand operating in this space: Personal transformation ranks as the second most-wanted outcome, right behind professional excellence. People don’t just want AI to do their jobs faster. They want it to help them become someone better.

And yet, by Anthropic’s own numbers, only about 6% of people feel AI has actually delivered on emotional support, personal guidance or a judgment-free space to think something through. Compare that to productivity, where roughly a third of respondents say AI has clearly delivered. People are asking AI for personal growth — and getting a faster spreadsheet instead.

That disconnect is an opportunity for brands. ChatGPT, Claude, Gemini, Copilot, Grok and Meta AI are all selling productivity first — smarter, quicker, better at doing things for you. That’s a compelling value proposition. But it’s a very different one from “this will make you a stronger version of yourself.” Right now, almost no major player is making that pitch.

And whoever solves for that outcome has a chance to define what “the good AI” looks like in the public imagination for years to come.

We’ve seen all this before

To understand why “the good AI” is such a challenge for the big players, we don’t have to look back very far. Just to Mark Zuckerberg and Facebook.

By most accounts, Zuckerberg didn’t create Facebook to fracture attention spans and feed people a steady diet of outrage. It set out to connect people — and then optimized, quarter over quarter, for the metric that paid the bills: engagement.

Unfortunately, engagement rewards whatever keeps a thumb moving, and outrage, comparison and dependency all keep a thumb moving better than human connection does. 

The impact of AI platforms, for good or ill, will be far greater. The technology is more intimate than a social feed ever was. It talks back, it remembers, it can sound like it cares.

Built the wrong way, optimized for engagement or usage minutes instead of a user’s best interests, it doesn’t just capture attention. It can quietly replace judgment. An AI that always has the answer, always agrees, always smooths the path — that’s not empowerment. That’s a more sophisticated version of the same trap, dressed up as help.

What the alternative looks like

Fortunately, there is a countertrend worth watching: AI built explicitly to build the user up rather than make itself indispensable. Instead of answering with the usual wall of advice and five-step plans, these tools ask questions — the way a good counselor or coach does — and let the person arrive at their own conclusions.

This movement is in its early stages and is mostly the territory of researchers and small players rather than the major labs. MIT Media Lab has built ImpactBench, which scores AI systems on whether real back-and-forth conversations with them leave a person better off or worse off. 

HumaneBench, built in response to real lawsuits over chatbots harming vulnerable users, tests whether AI protects a person’s well-being. Vesela’s Sovereign Human Benchmark goes further: Does the AI make a person stronger over time, building independence instead of reliance?

Optimizing for more than profits

None of this is coming out of the organizations building the most widely used AI products. But it’s aiming at something real: Everyone can measure engagement and retention; almost nobody has a working measure for whether an AI interaction left a person better off.

Whether any single company gets there first is a separate question from whether the goal is right. It is. “AI that makes you stronger, not more dependent” is a real answer to a real fear, and it’s the kind of stance a brand can actually own if it’s willing to build the whole product around it, not just bolt it on as an optional feature or tagline.

This is counterintuitive to how tech companies have learned to maximize profits. It’s essentially asking them to design friction back into the process where friction serves the person; a good coach doesn’t always make things easy. 

That’s a hard concept to sell in Silicon Valley or on Wall Street, but it’s not unheard of. Costco caps its margins instead of maximizing them. DuckDuckGo won’t build an ad profile on you. Signal won’t either, and doesn’t want your data in the first place. Basecamp has spent two decades rejecting growth-at-all-costs tactics. Each turned restraint into a business model. 

Nobody’s proven that can work in AI yet. Maybe some savvy, courageous company will make itself the first.

Fighting for a human-centered future

Easy answers are addictive. A tool that always tells you what to do, always agrees, always removes the friction of deciding for yourself — that’s a genuinely comfortable product to use, and comfortable products win in the short run. The human-centered approach asks more of the user. It asks them to earnestly grapple with a question instead of getting an answer handed to them.

The 81,000 people Anthropic talked to didn’t say they wanted an easier life. They said they wanted to become someone better. That’s a harder thing to build toward — but a much better future to fight for.

Long before AI showed up in most people’s lives, it already had a reputation. For a lot of people, that reputation is Skynet, the AI from the Terminator movies that decides to eliminate humans by starting a nuclear war. Our fears today are a bit more subtle: not that AI will destroy us, but that it’ll quietly take over people’s jobs and replace their judgment, one convenient answer at a time.

Anthropic recently published one of the largest studies of its kind — 81,000 people interviewed, conversationally, about what they actually want from AI. Here’s the finding that should matter to every brand operating in this space: Personal transformation ranks as the second most-wanted outcome, right behind professional excellence. People don’t just want AI to do their jobs faster. They want it to help them become someone better.

And yet, by Anthropic’s own numbers, only about 6% of people feel AI has actually delivered on emotional support, personal guidance or a judgment-free space to think something through. Compare that to productivity, where roughly a third of respondents say AI has clearly delivered. People are asking AI for personal growth — and getting a faster spreadsheet instead.