AI Is Making Us Faster—But Not Smarter (And Why That’s a Big Problem)

A rare stance on innovation amidst the AI productivity stampede.

Every problem today has the same answer: AI. It’s like a hammer that sees a world filled with nails. But what if we’re nailing the wrong thing at record speed? AI seems like the answer to everything, until it isn’t.

AI’s potential to speed up corporate productivity has stolen center stage. This obsession we have with productivity is ancient. And it’s the wrong measure of success. Though, we are clamoring for it more than ever in the race to adopt AI.

Everyone is trying to use AI to scale their operations. But are they simply scaling the same, tired output strategy of better, faster, and cheaper? I contend they are. Most companies are busy trying to figure out how they can use AI to solve everything. They aren’t thinking about if they should.

Let’s explore this deeper and see what, I believe, is a better way to capitalize on AI to fuel innovation.

The difference between efficient, productive, and effective.

Productive is better than efficient, but both are inferior to effective. And AI falls short on the effective front. To explain, let’s first define how I use these three terms.

• Efficiency is about the utilization of a finite resource to perform a task. These tasks are inputs that add up to a finished work product. If the resource is idle, it’s less efficient. Think of a factory optimizing to keep its workers busy. They want every dollar they pay the worker to apply to the worker doing work. They avoid idle workers at all costs.
• Productivity is about optimizing the production of work to maximize throughput. Here the efficient flow of the work to completion is crucial. You aim to keep the assembly line moving. Often, in practice, managers optimize both worker efficiency and productivity of the work. In other words, keep the workers busy and keep the work output flowing without pause.
• Effectiveness is about the results of what gets produced. Effectiveness has no relationship to efficiency and productivity. You can be efficient and productive and produce something that isn’t useful. And you can be inefficient and unproductive yet produce something extraordinary and impactful. If you focus on the wrong thing, productivity doesn’t matter. The efficiency of workers makes no difference. Everyone might as well take a nap.

So, which of these do you think AI can improve?

• AI is great at making more, but terrible at making right.
• AI maximizes output. Humans maximize impact. Both must be in concert.
• More speed, more output, more efficiency—none of it matters if we’re still building the wrong thing.

AI can produce faster. It can produce cheaper. But it can’t differentiate on the effectiveness of what it spits out. It can’t produce better.

But none of this stops our dogged pursuit of efficient productivity—keeping people busy. We give little thought to if it’s the right thing. This is a management puzzle we’ve been trying to solve (in vain) as long as humans have organized to achieve a goal.

How the AI chase is just the same age-old productivity chase.

Before AI, the answer to scaling productivity was to add more people and keep them busy flowing more work. But now, instead of people, AI offers the chance to scale with almost limitless capacity. Instead of people, It’s about adding GPUs with only energy as the limiter.

GPUs are cheaper than people. They don’t complain. And they work around the clock without tiring.

With AI, you produce faster inputs to feed faster output. And let’s assume the quality is good (this is highly questionable with today’s LLMs). You will get piles of adequate output from AI. But, remember, productive output is not effective output.

Let’s take the software product development space. 80% of all features produced are rarely or never used.^1 Guess what productive AI will achieve? It will help us produce even more piles of unnecessary features. Is this the right direction? That’s a definite, “No.” We will end up with more of that which was never needed.

I can almost hear a counter-argument: “But with AI, you could learn you are wrong faster.”

Well, this is true, but why haven’t we done this without AI? We don’t learn we are wrong early today. We haven’t improved our poor track record in software development in over 50 years. Why would it be any different with AI?

Most of us humans aren’t interested in learning we are wrong. Instead, we try to prove we’re right. And with uber-productive AI, we will deliver our biased ideas like we do today, even faster.

It’s better to find the right thing to do and then do it fast, than to only do things fast. But finding out the right thing to do is as much a blind spot for AI as it is for us.

AI can’t predict outcomes, it can only speed up productivity.

Is the right thing predictable with AI any more than without it?

If only the right thing were quantifiable and knowable upfront. Then, AI would have a chance to predict it and build the right thing the first time. But the effectiveness of a solution is reliant on factors not always quantifiable.

Take customer delight and adoption of what you produce. Customers are fickle, and their needs are fleeting. What they want today, they don’t want tomorrow. It’s complete unpredictability.

You must be in the right place at the right time to thread the needle and capture the hearts and minds of a target market.

This takes many small bets and pivots. Most bets won’t land. It requires a non-linear series of bets to find the perfect path to a customer. You need empathy and human judgment to uncover this path, one small step at a time. No data exists to tell you upfront what will work and what won’t.

And last time I checked, AI’s not that great at empathy and human judgment. It’s all about efficient productivity.

So, should productivity be the main target for AI? Is this good enough?

With AI, we can now produce the wrong thing exponentially faster than before. And we will unless we shift AI to target something better than productivity. AI needs to target innovation—alongside humans, rather than without us.

Innovation is the better target for AI, but it requires human touch and judgment.

Figuring out the right thing to build requires human interaction. Humans bring empathy, nuance, and taste. These things AI can’t do alone. Even artificial general intelligence won’t match the human here. Until we achieve artificial general human emotion, innovation will escape AI’s grasp.

Innovation is born of something new. It doesn’t mimic. It doesn’t copy. It doesn’t plagiarize. Yet, this is the basis of modern AI. It learns from that which already exists. Creation escapes it.

For example, take China and DeepSeek R1.

Look at their innovation, born from external constraints and paired with human ingenuity. They likely utilized AI to realize their idea faster. But their breakthrough required the human spirit, ignited by competition and imposed sanctions. They came up with the ideas. Then, they used efficiency and productivity to realize them faster. This was a marriage of the human and the machine to produce a novel AI model at a record pace.

True, DeepSeek might have sailed on the coattails of OpenAI’s model by learning off its answers. But it created striking innovation in its model. It has unmatched learning efficiency, low GPU processing needs, and breathtaking runtime speed. This required human ingenuity. No AI could have done this alone.

We have to make the goal of AI to help us innovate faster.

We need AI to help us increase the learning loop speed. Not to produce more bad ideas faster with peak productivity. But to innovate through faster experimentation.

If AI can help us place more bets with efficiency and productivity, we can try them out faster. We can learn faster. We can pivot fast, trashing what fails, keeping what works, and iterating what shows promise.

This is what fuels innovation. And it requires a human and AI working in concert. One or the other in isolation misses the opportunity in the union of both.

It’s the graceful innovation dance of the AI and the human that will win.

---

TL;DR

AI should enhance our human innovation, not be a solution for productivity nirvana.

AI is great at:

• Maximizing efficiency → Automating repetitive tasks.
• Boosting productivity → Producing more, faster, at scale.
• Failing at effectiveness → AI can’t decide what’s actually worth making.

With AI, we’re at risk of producing the wrong things faster than ever. Faster failure is still failure.

Don’t chase the AI productivity mirage and die of dehydration. Quench your thirst by finding what’s right to focus on through rapid AI experimentation.

Shift from a productivity focus to an innovation focus, with humans in the driver’s seat. Then, once you aim right, you can be productive as hell—building the innovations your customers need.

• Is AI helping or hurting innovation in your industry?
• How are you using AI—for scaling productivity or fueling innovation?
• AI can make us faster, but can it help us be smarter and innovate faster?

Let’s discuss. Drop your thoughts below.

---

➡️ Sign up for weekly insights like this. Learn about returning to the fundamentals that make product teams unstoppable. Join a community of like-minded professionals committed to achieving outcomes sooner.

References

1. Thomas, S. (2019). 2019 Feature Adoption Report. Pendo.io. Retrieved from https://go.pendo.io/rs/185-LQW-370/images/2019%20Feature%20Adoption%20Report%20Digital.pdf