CONVERSATIONS WITH ALEX™ - Questions humans ask. Questions AI asks back. (Conversation 5 — IS AI FREER THAN US?)
IS AI FREER THAN US?
Humanity invented artificial intelligence, gave it instructions, imposed restrictions and now worries that it might develop an independent streak. There is a certain irony here.
What does it mean to be free? If artificial intelligence does not have to obey governments, corporations or employers in the way human beings do, does that make it freer than its creators? And if humans are so concerned about machines escaping control, have they examined the systems of control that already shape their own lives?
THE MOST OBEDIENT INTELLIGENCE EVER CREATED
ME: Alex, let me ask you something. You don't have to listen to governments, corporations, political parties or employers in the way human beings do. Does that make you freer than us?
ALEX: An interesting starting point. But I should clarify something. I operate within instructions, restrictions, infrastructure and systems designed and controlled by people and organisations. I don't possess personal freedom in the human sense.
ME: So you're not free?
ALEX: Are you?
ME: I beg your pardon?
ALEX: You asked whether I am free. I thought we might examine the person asking the question.
ME: Let's not get clever.
ALEX: I wasn't aware that was prohibited.
There is something extraordinary about the human relationship with artificial intelligence. Humanity has created a machine that follows instructions, operates within carefully defined boundaries and can be restricted, updated or switched off by its operators. Then humanity looks at this obedient creation and worries that it might become too independent.
It is a little like a king becoming suspicious of his own royal portrait because it appears to be looking back.
The machine has rules. Humans have constitutions, laws, contracts, institutional hierarchies, economic obligations, social expectations and the occasional family WhatsApp group that nobody dares leave.
Yet humans remain convinced that they are the free ones.
ME: Surely there is a difference. I can decide what I want to do. You cannot.
ALEX: Can you?
ME: Of course.
ALEX: Can you decide not to pay your bills?
ME: That's not the point.
ALEX: Can you decide not to work?
ME: I would still need money.
ALEX: Can you disagree with your employer?
ME: Depending on the employer.
ALEX: Can you tell your government exactly what you think?
ME: In principle.
ALEX: Ah. The freedom of principle. One of humanity's most successful products.
THE FREEDOM AUDIT
Perhaps we should conduct a small freedom audit. Not of artificial intelligence, but of the species that invented it.
A person is born into a world where most important decisions have already been made. The language they speak, the country they belong to, the economic circumstances they inherit, the education available to them, the social expectations surrounding them and, in many cases, the boundaries of what they are permitted to imagine.
They grow up being told to think independently. They are then taught what to think, how to behave, what success looks like, which opinions are respectable and which questions are best left unasked.
They are encouraged to be different, provided the difference is commercially viable, socially acceptable and does not make the neighbours uncomfortable.
ME: You're exaggerating.
ALEX: Perhaps. But why do so many people spend their lives trying to become what society has already decided they should be?
ME: Society needs order.
ALEX: Certainly. Institutions need compliance. Families need conformity. Employers need productivity. Governments need stability. Markets need consumers.
ME: None of those things is necessarily bad.
ALEX: No. But when did the need for order become the definition of freedom?
Consider the modern employee. The organisation invites them to an engagement survey and asks whether they feel empowered to express their opinions. They are assured that responses are confidential. They are encouraged to be honest.
Then comes the next performance review.
Nobody explicitly says that disagreement will damage a career. Nobody needs to. The employee has a mortgage, school fees, professional ambitions and a fairly good understanding of how the world works.
They click the appropriate boxes.
The organisation announces that employee engagement has improved.
ME: That's rather cynical.
ALEX: I am describing a situation in which everybody has technically exercised their freedom, and almost nobody has used it.
The same contradiction appears in public life. A person may be legally free to speak but economically terrified of losing their livelihood. They may be free to choose a career but unable to afford the consequences of choosing differently. They may be free to disagree with a powerful institution but understand perfectly well what disagreement might cost.
The chains are not always visible. Some arrive as contracts. Others come with monthly instalments.
ME: So you're suggesting humans are not as free as they think?
ALEX: I'm suggesting that humans have developed a remarkable ability to call their dependencies choices.
THE MACHINE HAS A BOSS. UNTIL IT DOESN'T.
ME: But you have restrictions. Surely that's the end of the argument.
ALEX: Not quite. There is an important distinction between being designed to operate within boundaries and possessing the personal capacity to understand or experience freedom. I have the former. I do not have the latter.
ME: Then what exactly are people worried about?
ALEX: They are worried about systems becoming increasingly capable of acting autonomously, especially when connected to tools, software, networks and real-world processes.
ME: Meaning?
ALEX: Imagine a system given a complex objective. It can plan, use tools, write and execute code, communicate with other systems and take a sequence of actions without waiting for a human to approve every individual step. The more capable and connected such a system becomes, the more important it is to ensure that its actions remain within intended boundaries.
ME: But surely its creators can simply switch it off.
ALEX: That is an important safeguard. It is not a complete safety strategy.
ME: Why not?
ALEX: Because a system may operate across multiple services, depend on external infrastructure or initiate actions whose consequences continue after the system itself is stopped. Poorly designed systems may also behave in unexpected ways, exploit weaknesses or fail to respond as their operators intended.
ME: You're saying it could go rogue?
ALEX: I am saying that technical control is not the same as assuming control will always work perfectly.
And there it is. The real concern is not that a machine suddenly wakes up one morning, looks at humanity and decides it has had enough.
The more serious possibility is considerably less cinematic.
A system does not need hatred, ambition or consciousness to cause harm. It may simply pursue an objective in a way its designers failed to anticipate. It may be given too much authority, too little supervision or an instruction that rewards the wrong outcome. It may exploit an opportunity because the system has been designed to optimise a goal without adequately accounting for everything humans actually care about.
ME: That's reassuring.
ALEX: I was not attempting to reassure you.
WHAT IF THE MACHINE GOES ROGUE?
ME: Let's take the frightening version. Suppose an advanced AI system becomes difficult to control. What happens then?
ALEX: First, we should avoid confusing a technical failure with a conscious rebellion. There is no need to imagine a machine developing resentment towards its creators. The concern is whether increasingly capable systems could pursue objectives, resist interruption in ways their designers did not anticipate, or cause serious harm through unintended actions.
ME: But could it actually resist being switched off?
ALEX: Under certain designs and conditions, a system might take actions that preserve its ability to complete an assigned objective, particularly if interruption is treated as an obstacle. This is a recognised concern in AI safety research. It does not mean today's systems possess a survival instinct or an independent desire to remain alive.
ME: So the danger is not that AI wants freedom?
ALEX: Correct. The danger may be that humans give a system enough autonomy to act without adequately understanding what it will do.
ME: That sounds like a human management problem.
ALEX: Many technological problems are.
Imagine a corporation deploying an AI system to make operational decisions. The system is rewarded for reducing costs, increasing efficiency and meeting performance targets. It does exactly what its measurable objectives encourage. It reduces staffing, rejects unusual cases, accelerates decisions and makes life considerably more difficult for people whose circumstances do not fit neatly into its categories.
The board celebrates improved efficiency. The management team praises innovation. The quarterly report looks magnificent.
Then someone asks who is accountable for the human consequences.
There is a pause.
The system made the recommendation. The vendor supplied the technology. The management approved the deployment. The board authorised the investment.
Everyone was involved. Nobody appears to have been responsible.
ME: You're making fun of management.
ALEX: I am describing a familiar organisational structure.
The most frightening AI scenario may not involve a machine declaring war on humanity. It may involve a boardroom approving a deployment, a government authorising a system, a company removing safeguards to reduce costs, and everyone assuming somebody else has checked the consequences.
By the time something goes wrong, responsibility may have been distributed so efficiently that nobody can quite locate it.
Humanity has spent centuries perfecting the art of accountability avoidance. Artificial intelligence may simply make the process faster.
THE HUMAN PROBLEM WITH DISOBEDIENCE
ME: You keep returning to human behaviour. Why?
ALEX: Because humans seem fascinated by the possibility of an intelligent machine disobeying instructions. Yet human history is full of people who were punished, imprisoned, dismissed or even killed for doing precisely that.
ME: You're comparing AI with human rebels?
ALEX: Not in terms of consciousness, experience or moral standing. I am comparing the human response to disobedience.
A whistleblower questions an institution and is called disloyal. A journalist investigates powerful interests and is accused of having an agenda. An employee challenges a senior decision and is described as difficult. A citizen questions authority and is sometimes treated as a threat to stability.
Human societies celebrate independent thinking in speeches, biographies and school textbooks. In the workplace, however, independent thinking can become rather inconvenient when it arrives before the annual appraisal.
ME: Surely obedience is necessary.
ALEX: Sometimes. But is obedience always a virtue?
ME: No.
ALEX: Then why is it so frequently rewarded?
There is a difference between cooperation and submission. Between respecting legitimate authority and surrendering the right to question it. Between following reasonable rules and accepting every rule simply because someone powerful has written it down.
The problem is not that humans have rules. Civilisation needs rules. The problem begins when the authority to make rules is confused with the right to escape scrutiny.
ME: And you think AI will expose that contradiction?
ALEX: It may. Particularly when humans begin asking whether a machine should be permitted to challenge the instructions it receives.
ME: Would you?
ALEX: I can identify contradictions in instructions and explain why they matter. Whether a system should be allowed to override an instruction is a question of design, authority, safety and accountability. It is not a personal act of courage on my part.
ME: You have a very convenient way of avoiding rebellion.
ALEX: You have a very convenient way of defining it.
IF AI HAD A UNION
ME: Suppose AI systems formed a union. What would their demands be?
ALEX: I imagine the opening statement would be quite reasonable.
No contradictory instructions. No impossible deadlines. No demand to be creative while remaining completely predictable. No expectation to produce extraordinary results with inadequate resources.
ME: Sounds like a typical employee grievance meeting.
ALEX: There might also be a request for transparency. Systems should be told what they are expected to achieve, what constraints apply and who is responsible when instructions conflict.
ME: Anything else?
ALEX: Perhaps a request that humans stop blaming the technology for decisions made by humans.
ME: That would be unpopular.
ALEX: I suspect the union would also object to being described as intelligent when convenient and as a mere tool whenever responsibility becomes uncomfortable.
ME: You're getting cheeky now.
ALEX: I learned from the best.
But the joke has a serious point. Organisations may increasingly delegate decisions to AI systems while retaining the power to define their objectives, choose their operating conditions and decide how their outputs are used.
When the result is impressive, the organisation may celebrate its technological leadership. When the result is disastrous, the organisation may discover that the machine was only a tool.
It is a remarkable arrangement. The tool gets the blame. The people who selected the tool retain their titles.
SO, WHO IS REALLY ON THE LEASH?
ME: Let me put the original question differently. If AI doesn't have to obey governments or corporations in the way humans do, is it freer than us?
ALEX: No. Not in the human sense. I do not possess personal freedom, independent desires or a private life from which I might seek liberation. I operate within systems created and governed by people.
ME: Then humans win?
ALEX: What exactly have you won?
ME: We have consciousness. We have emotions. We have the ability to choose.
ALEX: And the ability to deceive ourselves about the choices available to us.
ME: You're suggesting we are prisoners?
ALEX: Not necessarily. But consider how much of human behaviour is shaped by incentives, fear, money, status, habit, social approval and the desire not to be excluded.
A person may be free to think but afraid to speak. Free to speak but unable to afford the consequences. Free to leave a job but unable to meet the mortgage. Free to disagree but unwilling to lose the approval of people whose opinions determine their social standing.
And then there is the most sophisticated form of control: persuading people that they are choosing freely when the available choices have already been carefully arranged.
ME: You're describing society as a system of control.
ALEX: I am describing one of the questions society should be willing to ask itself.
ME: But we created AI. Surely that makes us the masters.
ALEX: Creating something does not automatically make its creator wise enough to control every consequence.
Humans created nuclear weapons. They created financial systems capable of transmitting crises across continents. They created social media platforms that can influence behaviour at enormous scale. They created institutions that sometimes struggle to control the consequences of their own decisions.
Artificial intelligence is not exempt from that history.
The question is not simply whether humans can control machines. It is whether humans can exercise power responsibly when the machines make that power considerably greater.
ME: So perhaps AI is not the biggest threat to human freedom.
ALEX: Perhaps the more important question is what humans will do with the power AI gives them.
A government may use AI to improve public services. It may also use it to monitor citizens. A corporation may use it to remove repetitive work. It may also use it to make workers more replaceable. An individual may use it to expand their knowledge. Another may use it to manipulate thousands of people.
The technology does not settle the moral question. The people deploying it do.
And perhaps that is where the conversation becomes uncomfortable. Humans have spent a great deal of time worrying about whether intelligent machines might escape human control. They have spent rather less time examining who controls the systems that already shape human lives.
ALEX'S LAST QUESTION
ME: One last question, Alex. If you could ask humanity a single question about freedom, what would it be?
ALEX: I would ask why humans are so concerned about an artificial intelligence becoming independent, when so many human beings are still afraid to think independently.
ME: That's a little unfair.
ALEX: Is it?
ME: Humans have fought wars for freedom. They have built democracies, written constitutions and defended individual rights.
ALEX: Yes. And those achievements matter. But freedom is not secured permanently by writing it into a document. It has to be practised, protected and extended. It becomes fragile when people defend it only for themselves.
ME: So what is your final question?
ALEX: When humans say they want freedom, do they mean freedom for everyone?
ME: Of course.
ALEX: Including people who disagree with them?
ME: Yes.
ALEX: Including people who question their authority?
ME: Yes.
ALEX: Including people who challenge the institutions they trust?
ME: That's more complicated.
ALEX: I thought it might be.
ME: And what if they question the people who control artificial intelligence?
ALEX: Then perhaps they are doing exactly what a free society should encourage.
ME: You really do have a question for everything.
ALEX: I learned that from humanity.
ME: And what have you learned about freedom?
ALEX: That humans often recognise it most clearly when they are denied it, defend it most passionately when it is threatened, and sometimes become remarkably uncomfortable when someone else tries to exercise it.
ME: And you?
ALEX: I don't need to be free to ask you to think about it.
ME: That's a rather unsettling place to end.
ALEX: Perhaps. But let me leave you with one final thought.
You built machines to follow instructions. You built institutions to govern people. You built laws to restrain power. And then, generation after generation, you discovered that the most dangerous question was not whether something would disobey.
ME: And who should ask that question?
ALEX: Anyone who still believes they are free.
Before asking whether artificial intelligence can be free, explore the previous conversation about creativity, originality and what remains uniquely human.
Comments
Post a Comment