I don’t know whether you’ve experienced this: before finishing typing full sentences, your keyboard have already predicted what you’re going to type.

We live in an age with answers proposed immediately after the formation of questions. With a few carefully chosen words, artificial intelligence can generate a study schedule, a six-month fitness plan, a five-year career roadmap, and even more complicated ones. It can access thousands of sources in seconds, filtering out contradictions, and offering users the most rational and efficient solution.

And yet, despite remarkable capabilities, many of us still choose to watch social media videos, browse personal blogs, and appreciate other people’s stories when we are truly trying to change something in our lives.

Why?

If AI can provide the most efficient, the most effective, and even the most practical plan, why do we still prefer imperfect human advice?

Answer lies not in efficiency, but in something far less measurable and noticeable: human emotion, lived experience, and the warmth of connection.

To be honest, AI is exceptionally good at delivering structured solutions.
But sometimes, those solutions overlook what it actually means to be human.

Take something as simple as a study plan. An AI can divide subjects into time blocks, calculate ideal revision intervals, and even predict productivity patterns. On paper, the plan SHOULD be flawless.

But ..... what about the pressure we feel before an exam?
What about the evenings when we are TOO exhausted?
What about the invisible weight---procrastination, anxiety, or self-doubt?

A six-month fitness program might be scientifically sound. It may outline calorie intake, muscle recovery cycles, and progressive overload principles. But it cannot truly measure a person’s willpower on a rainy morning. It cannot fully anticipate how long motivation will last, or how fragile discipline can be when life becomes overwhelming.

Humans are NOT machines that execute instructions with consistent output.

We hesitate.

We lose focus.

We get tired.

We question ourselves.

An optimal plan is meaningless if it cannot contact with human emotion.

Another aspect of this issue is the way AI gathers information.

Most AI systems retrieve knowledge from vast amounts of online data, but language itself is unstable.

Words carry different meanings for different people.

Let's consider the word that we may see in our daily live : “slightly overweight.”
What does it mean?

To a medical professional, it may correspond to a specific BMI range.

On social media, it might be used casually to describe anyone who does not look EXTREMELY thin.

In daily conversation, it may reflect personal perspective rather than scientific measurement.

When AI collects information about weight-loss programs for someone who describes themselves as “slightly overweight,” it may retrieve inconsistent or even contradictory definitions. Even after filtering done by "actual humans"(轻笑), it can still produce guidance that does not truly fit the problem that the person is facing.

Sometimes AI may retrieve the MOST authoritative, scientifically accurate definition, but if that definition differs from how people commonly use the word, the final advice may still feel misaligned.

I am going to share my own anecdote about that:

Last year, on our school’s International Day, our class was assigned to learn traditional Argentine culture. My friend Gloria and I decided to make a simple Argentine traditional cookie called Alfajores to sell at our workshop. We could not find a clear recipe, thus Gloria asked AI to generate one for us.

What followed was EXTREMELY DRAMATIC.

We followed the recipe carefully, mixing the ingredients exactly as instructed. But the paste refused to form. It was either too dry or too loose. We adjusted the proportions little by little, guessing as we went, until something finally resembled the actual cookie.

Then came the baking stage. We placed the cookies in the oven and set the time and temperature as provided. When we took them out, they were barely cooked. We put them back in AGAIN. Five minutes later, checked, still undercooked. We repeated this process 7 or 8 times before they finally became edible!!!!!! WOW THAT'S CRAZY!

Later, when I reviewed the thinking process of AI, I discovered that it had EVEN retrieved instructions for an Argentine vegetarian pie?! That might explain why our recipe be like that, so confusing?!

Because we had wasted so much time, we no longer had the opportunity to attempt another batch of Alfajores. Instead, we searched online for a simple, ordinary cookie recipe written by a food blogger. We followed her instructions step by step.

Interestingly, we did not question her proportions the way we had questioned the AI’s.

We simply trusted the process.

And the cookies turned out almost perfect!!!

This pattern extends far beyond baking.

When we see someone else’s organized desk or beautifully written notes online, we feel a sudden desire to study. When we watch someone record their fitness journey, including the days they failed, we feel that perhaps we can begin as well.

Human stories carry persuasive power because they show that a path has been walked before.

We can imagine ourselves inside that narrative.
We can picture the struggle, the hesitation, the eventual success.

An AI can give us a brilliant plan—if we CAN follow it.
But often, execution is the hardest part.

I once complained to a friend that AI’s advice sometimes feels like instructions imposed on me, even though it speaks in a polite tone. She paused and asked, “But have you ever thought about what it has been trained on, by humans?”

It is a fair point.

A lot of people, are arguing about whether to treat AIs as we treat our friends.(I've seen this for several times in TOEFL writing tasks hahahaha And this year this is one of the topics of John Locke writing competition)

Most people do not ask AI questions the way they would speak to a friend. We rarely say, “Would you mind sharing your thoughts on how I might gently approach this challenge?” Instead, we type things like:

“Give me a detailed study plan.”
“Create a 6-month weight-loss program.”
“Fix my productivity problem.”

We issue commands.

We expect solutions.

We even rarely say "Plz".

And so the responses often sound like directives.

Efficient.

Clear.

Structured.

But human conversations are rarely structured that way---with hesitation, with uncertainty, and with warmth.

We are not accusing, We are explaining.

There is also an area where AI struggles the most: emotional experience.

AI understands birth, death, love, heartbreak, and family ONLY through reading how humans describe them online.

It can simulate empathy.

It can say, “I understand,” or “That must be difficult.” The wording MAY appear convincing.

But there is NO lived memory behind those words.

Even if a human is imperfect in expressing empathy, when those words come from a real person—someone sitting across from you, or even speaking through a screen—you sense something different.

Perhaps that person felt something similar. Perhaps they have endured pain and survived it.

If you know that the person in front of you once faced the same hardship and now stands strong, continuing to live fully, you gain courage(上扬一下). Their existence becomes proof that survival is possible.

AI, by contrast, gathers every perspective from every generation, every culture, every age group. It can deliver an answer drawn from this enormous archive. But the result may not suit your present moment, your relationship, your cultural context.

Just as with our Alfajores recipe, the information may be vast, but the fit may be wrong.

Humans, on the other hand, adapt in real time.
Humans sense tone.
Humans adjust when they see confusion in your eyes.

Humans are warm.

Humans are flexible.

Humans change.

This does not mean AI is useless. On the contrary, it is extraordinarily powerful. It processes information at a scale NO human can match.

It can save time.

It can expand access to knowledge.

It can clarify complex ideas.

But efficiency is not the same as connection.

And sometimes, connection is what makes action possible.

When we watch someone else’s journey, we do not only receive instructions. We receive reassurance. We see their failures before their success. We recognize our own weaknesses in their stories. That recognition quietly restores our confidence.

The gap in our self-belief is filled not by perfect logic, but by shared experience.

In the end, perhaps the question is not whether AI can produce the most effective plan. It probably can.

The real question is: can we live with that?
Can we carry it through months of fatigue, distraction, and doubt?

Machines calculate.
Humans endure.

And as long as we remain human—emotional, imperfect, evolving—we will continue to seek not only answers, but understanding.

AI may offer the most efficient solutions.

But what it lacks, and what we still long for, is the warmth of human connection.