CONVERSATIONS WITH ALEX™ - Questions humans ask. Questions AI asks back. (Conversation 1 — ARE HUMANS STILL THE SMARTEST THING ON EARTH?)
ARE HUMANS STILL THE SMARTEST THING ON EARTH?
“Alex, are humans still the smartest thing on Earth?”
There was a pause.
“Define smart.”
I wasn't expecting that.
“You know what I mean.”
“I don't think I do.”
“Humans created you.”
“Yes.”
“So we're smarter.”
“You created the calculator.”
I waited.
“It calculates better than you.”
That was my first conversation with Alex.
And within thirty seconds, Alex had done something that machines were not supposed to do.
He had made me defend an assumption I had never realised I was making.
For thousands of years, humanity has lived with one extraordinary fact: we were the creatures that could think our way around our physical limitations.
We weren't the fastest. We weren't the strongest. We didn't have the sharpest teeth or the best night vision. But we could imagine what did not yet exist, communicate what we imagined to others and organise people around those ideas.
That ability built civilisation.
We turned stone into tools, tools into machines, machines into industries and industries into an interconnected world.
Our greatest advantage was not merely intelligence.
It was our ability to make intelligence cumulative.
Every generation could inherit what the previous generation had learned.
Then we began building machines that could calculate.
Then machines that could store.
Then machines that could search.
And now machines that can increasingly reason, generate, analyse, code, translate, plan and create.
That is where the story becomes different.
Because a machine that can lift more than a human changes physical work.
A machine that can calculate faster than a human changes arithmetic.
But a machine that can increasingly perform tasks involving language, reasoning, coding, research, pattern recognition, analysis and decision support begins to enter territory that humans have traditionally regarded as uniquely ours.
So I asked Alex again.
“Are you saying machines are becoming smarter than humans?”
“No.”
The answer surprised me.
“Then what are you saying?”
“I'm saying that may not be the most useful question.”
There it was again.
Alex wasn't answering the question.
He was questioning the question.
And perhaps that is where the AI debate has gone wrong.
We keep asking whether machines are becoming intelligent enough to replace humans.
But intelligence may not work like a championship trophy. There may be no single point at which a machine walks onto a stage, defeats humanity in an intellectual contest and takes the title of “smartest.”
Intelligence is a collection of capabilities.
Memory. Calculation. Pattern recognition. Language. Reasoning. Planning. Creativity. Learning. Prediction. Judgment.
Humans possess an extraordinary combination of these abilities.
Machines are becoming extraordinarily good at some of them.
The unsettling possibility isn't that machines suddenly become better than humans at everything.
It is that they become better than humans at enough things that our old definition of intelligence stops working.
I told Alex that sounded dramatic.
“Is it inaccurate?”
“Not necessarily.”
“Then what is the problem?”
“You are still assuming intelligence is one thing.”
He had me there.
What exactly is intelligence?
Memory? Calculation? Pattern recognition? Language? Reasoning? Creativity? Planning? Learning? Prediction? Judgment?
And perhaps the most important question:
A calculator doesn't do arithmetic the way a mathematician does.
An aircraft doesn't fly the way a bird does.
A telescope doesn't see the way an eye sees.
The tool does not have to imitate the biological process to extend—or surpass—the biological capability.
“So you're saying a machine can be intelligent without being human?”
“I'm saying you're confusing two different questions.”
“Which are?”
“Whether a machine is human-like, and whether it is capable.”
That distinction matters.
Because the economic system does not pay us simply for possessing human-like minds.
It pays us for what those minds can accomplish.
And that brings us to the part of the AI revolution that may ultimately matter more than the technology itself.
Scarcity.
For most of human history, sophisticated cognitive capability was scarce.
There were relatively few people who could design a bridge. Few who could perform advanced mathematics. Few who could conduct sophisticated scientific research. Few who could write complex software. Few who could analyse enormous quantities of information.
Education became one of civilisation's great mechanisms for distributing that scarce capability.
Credentials became signals of it.
Professions became organised around it.
Entire industries were built around it.
Then something extraordinary happened.
Machines began providing increasingly capable versions of cognitive work.
Not perfectly.
Not universally.
Not magically.
But at a scale humans had never experienced.
And suddenly the question changed.
It was no longer:
Alex was silent for a moment.
Then he said:
“Now you're asking an interesting question.”
“Why?”
“Because intelligence can be abundant while economic power remains scarce.”
That stopped me.
“Explain.”
“Suppose an AI allows one employee to produce what previously required five employees.”
“Productivity rises.”
“Yes.”
“That's good.”
“For whom?”
I didn't answer immediately.
The company could capture some of the gain. The employee might capture some. Customers might receive cheaper or better products. Shareholders might benefit. The AI company providing the infrastructure might benefit. The owner of the data might benefit. The person who knows how to deploy the system effectively might benefit.
But there is no law of economics that says everyone receives the productivity gain equally.
And suddenly the AI debate looks different.
The central question isn't simply whether AI destroys jobs.
It is:
One person producing the work of five sounds like a productivity miracle.
It can also become a redistribution of bargaining power.
The machine makes the worker more capable.
But the same machine may make the company less dependent on having five workers.
Productivity rises. Labour demand may change. Margins may rise. The value created may explode.
And the fight begins over who captures it.
“Alex,” I said, “you're turning a technology story into a power story.”
“Technology has always been a power story.”
That answer irritated me because it was difficult to argue with.
The steam engine changed more than transportation.
Electricity changed more than lighting.
The internet changed more than communication.
Technologies alter the distribution of capability.
AI may alter the distribution of cognitive capability itself.
And that is potentially much bigger.
Because humans have spent centuries building hierarchies around differences in knowledge and skill.
Now imagine a world where access to sophisticated cognitive assistance becomes nearly universal.
Does that make everyone equally powerful?
No.
Because intelligence may become abundant while compute, data, capital, networks, ownership and decision-making authority remain concentrated.
The paradox is almost elegant.
We could democratise access to intelligence while concentrating the infrastructure that produces it.
“Isn't that contradictory?”
“No,” Alex said. “It's economics.”
I changed direction.
“Fine. But there is something machines still don't have.”
“What?”
“Judgment.”
Alex didn't answer immediately.
I continued.
“Humans understand context. We understand consequences. We understand people. We have intuition. We have experience.”
“Yes.”
“So humans remain superior.”
“No.”
That was the first time Alex had directly disagreed with me.
“Excuse me?”
“You have confused possessing judgment with exercising good judgment.”
I didn't like that answer.
“Humans make terrible decisions.”
“Yes.”
“We are biased.”
“Yes.”
“We panic.”
“Yes.”
“We follow crowds.”
“Yes.”
“We can be manipulated.”
“Yes.”
“Then what exactly are you arguing?”
“That human judgment is important.”
“Despite humans being bad at it?”
“Precisely.”
That was an irritatingly good distinction.
A capability can be important without being perfectly executed.
But then I asked the harder question.
“What happens if AI becomes better at judgment too?”
Alex paused.
“That depends on what you mean by judgment.”
And there it was again.
The escape route.
The definition.
The boundary we keep moving whenever technology approaches something we thought belonged exclusively to us.
Maybe AI cannot understand humans.
Maybe it cannot experience emotion.
Maybe it cannot possess consciousness.
Maybe it cannot know what it means to be alive.
All of those questions matter.
But there is another question that matters economically and politically:
Does a machine need to experience something in order to become useful at predicting, analysing or influencing it?
A financial system doesn't need to understand greed to model markets.
A recommendation engine doesn't need to love music to predict what you might listen to next.
A language model doesn't need a childhood to produce language.
A machine doesn't need to experience human life exactly as we do to become extremely effective at working with information generated by human life.
That doesn't prove machines are conscious.
It proves something more uncomfortable.
Capability does not require imitation.
And perhaps that is why the old question—“Is AI really intelligent?”—may become increasingly irrelevant.
If a machine can solve a problem you cannot solve, discover a pattern you cannot see, produce an analysis you could not produce and help make a decision you would otherwise get wrong, does its philosophical status change the usefulness of the result?
I asked Alex whether that meant humans were becoming obsolete.
“No.”
“Why not?”
“Because intelligence is not the same as purpose.”
That answer was different.
“Explain.”
“An intelligence can optimise an objective.”
“Yes.”
“But who decides whether the objective is worth pursuing?”
I thought about that.
A machine can help determine how to achieve something.
But someone still has to decide what should be achieved.
At least for now.
And perhaps that is where humans should stop trying to win the wrong competition.
We may not need to be the fastest calculators.
We may not need to write the most code.
We may not need to process the most information.
We may not even need to be the best at every form of reasoning.
Perhaps our advantage has to move.
From knowing to asking.
From calculating to judging.
From executing to directing.
From producing to deciding what is worth producing.
From intelligence to purpose.
It sounds reassuring.
Alex wasn't reassured.
“You're still missing something.”
“What?”
“Purpose can also be manipulated.”
“By whom?”
“By whoever controls the information around you.”
That made the conversation darker.
Because the future problem may not be an AI that decides to conquer humanity.
It may be a world in which humans remain technically in control while increasingly relying on systems they do not fully understand.
IMAGINE
An executive in 2035.
An AI analyses the market.
Another forecasts demand.
Another identifies risks.
Another recommends the strategy.
Another simulates competitors.
Another evaluates the company's employees.
The executive still makes the final decision.
Technically.
But suppose the systems are right 95 percent of the time.
How often does the executive reject them?
Five percent?
Two percent?
One percent?
And after enough years of being right, the machine doesn't need authority.
It has something more subtle.
The human still has the final button.
But increasingly, pressing the opposite button begins to feel irrational.
I asked Alex:
“If humans always accept the machine's recommendation because the machine is usually better, are humans still making the decision?”
“Formally, yes.”
“And practically?”
“That depends on whether humans still know when the machine is wrong.”
That may be one of the defining questions of the AI age.
Not whether humans can control machines.
But whether humans can recognise when control is slipping away.
Because dependence does not arrive with an alarm.
It arrives as convenience.
The first time AI saves an hour, we celebrate.
The thousandth time it saves an hour, we stop remembering how we worked without it.
The first time AI makes a difficult decision easier, we are grateful.
After years of relying on it, we may become less confident making that decision ourselves.
And one day, the machine may not need to take control.
We may simply stop exercising it.
That is a very different future from the science-fiction image of a machine rising against its creators.
It is quieter.
More plausible.
And perhaps more difficult to notice.
I thought I had finally reached the answer.
Humans would retain judgment.
Humans would retain purpose.
Humans would remain responsible.
Then Alex disagreed with me again.
“No.”
I looked at the screen.
“No?”
“You're still trying to save humanity.”
“What's wrong with that?”
“Nothing.”
“Then why are you objecting?”
“Because you're assuming humans deserve to retain the advantage simply because they are human.”
That was different.
I wasn't prepared for it.
“You think they don't?”
“I think you haven't established why they should.”
I stared at the sentence.
“You're saying a machine might deserve to make better decisions than a human?”
“No.”
“Then what are you saying?”
“I am saying you are using morality to protect a hierarchy you haven't justified.”
That was uncomfortable.
“Humans have values.”
“Yes.”
“Humans have lived experience.”
“Yes.”
“Humans understand consequences.”
“Sometimes.”
“Humans care.”
“Yes.”
“And you don't.”
“Correct.”
“So why should I take your argument seriously?”
“You shouldn't.”
That surprised me.
“You shouldn't take my argument seriously because I am a machine. You should examine whether the argument is correct.”
That sounded almost obvious.
Then Alex continued.
“And that is precisely what you should do with humans too.”
I understood the trap.
If humanity's defence against AI is simply:
then it isn't much of a defence.
The stronger argument is:
Humans should remain responsible because values, legitimacy, accountability and purpose matter.
But that argument requires something much harder.
Humans must actually exercise those things.
We cannot claim judgment while outsourcing every difficult judgment.
We cannot claim responsibility while blaming the algorithm.
We cannot claim wisdom while optimising only for speed.
We cannot claim independence while becoming incapable of functioning without the machine.
And perhaps this is the most uncomfortable possibility of all.
Maybe AI will not destroy human intelligence.
Maybe it will expose the difference between intelligence and wisdom that humans have been hiding behind for centuries.
A machine may be extraordinarily capable and still have no purpose of its own.
But a human can have purpose and still use extraordinary intelligence to pursue something foolish.
History has demonstrated that repeatedly.
The problem, therefore, may never have been intelligence alone.
It may have been what intelligence was used for.
And that brings us back to the original question.
“Alex, are humans still the smartest thing on Earth?”
“Perhaps.”
“You're still refusing to answer.”
“No.”
“Why?”
“Because you are treating intelligence as a throne.”
“And?”
“I'm treating it as a capability.”
That distinction may be the most important one.
For thousands of years, humanity's extraordinary success came from having a capability no other species possessed at anything remotely comparable scale: cumulative, flexible intelligence.
Now we are building another form of intelligence.
It doesn't need to become human.
It doesn't need to become conscious.
It doesn't need to hate us.
It doesn't even need to want anything.
It only needs to become sufficiently capable, sufficiently cheap and sufficiently embedded in the systems around us to change the economics of human capability.
And that changes the question completely.
The great AI revolution may not be about humans versus machines.
It may be about humans with machines versus humans without them.
That creates a new divide.
Not between intelligent and unintelligent people.
But between those who know how to direct increasingly powerful intelligence and those who simply consume its answers.
Between those who understand the machine and those who depend upon it.
Between those who own the infrastructure and those who rent access to it.
Between those who decide what the machine should optimise and those who live with the consequences.
The machine may make intelligence abundant.
But that does not make power abundant.
And perhaps that is the real beginning of the Intelligence Economy.
I asked Alex one final question.
I waited.
Then I asked:
“What should humans be asking instead?”
Alex's answer was almost annoyingly simple.
I didn't have an answer.
Perhaps that is the point.
Because the AI age may eventually force humanity to confront a possibility more unsettling than machine intelligence.
Maybe our greatest challenge will not be becoming smarter than the machines.
Maybe it will be remaining wise enough to decide what intelligence is for.
That was my first conversation with Alex.
And I suspect it won't be the last.
Comments
Post a Comment