Yes, in that context "voice" almost certainly means the style, personality, and conversational behavior of the AI rather than an audio voice.
AI companies have unfortunately overloaded the word. It can mean:
-
Audio voice — the actual spoken sound, accent, cadence, male/female voice, etc.
-
Writing voice — the personality conveyed through language: formal, chatty, humorous, scholarly, concise, enthusiastic, and so on.
A button labeled "Meet the new voice" is often marketing language for "meet the new model personality" or "experience the new way it interacts," even if no sound is involved.
The confusion is made worse because ChatGPT also has a genuine voice mode where "voice" really does mean speech.
As a language enthusiast, you might appreciate that this is a fairly recent semantic expansion. Historically, a writer's "voice" was already a well-established literary term long before talking computers existed. Marketing departments have simply borrowed that meaning and mixed it with the audio one.
So your interpretation—that it refers to personality rather than sound—is very likely correct unless the button appears specifically within a speech or audio settings area.
That's a classic political philosophy problem, and one that has occupied thinkers from Aristotle to Mill to modern constitutional theorists.
A useful starting point is that a state dedicated to the well-being, safety, prosperity, happiness, and liberty of its citizens has multiple goals that sometimes conflict. Liberty pushes toward allowing conduct; safety and well-being push toward restricting some conduct. The question becomes: under what conditions does the restriction become justified?
One possible structure is to imagine progressively stronger justifications.
-
Direct harm to others. This is the strongest and least controversial case. Murder, assault, theft, fraud, kidnapping, and similar acts violate the rights or interests of others. Criminalization is generally justified because liberty itself is threatened if people cannot be protected from such harms.
-
Substantial risk of harm to others. Conduct that is not itself harmful but creates a serious risk may also be criminalized. Drunk driving is the standard example. The driver may arrive home safely, but the risk imposed on others is sufficiently great that society need not wait for an actual collision.
-
Protection of institutions necessary for liberty and prosperity. Perjury, bribery, counterfeiting, espionage, and tax evasion may not always produce an immediately identifiable victim, yet they undermine systems on which everyone depends. A state may criminalize conduct that threatens the functioning of courts, markets, elections, or national defense.
-
Protection of those unable to give meaningful consent. Laws concerning children, severe cognitive impairment, or certain forms of exploitation rest on the view that liberty sometimes requires protecting people who cannot adequately protect themselves.
The more difficult questions arise when no one except the actor is harmed.
-
Paternalism. May the state criminalize conduct because it is bad for the actor himself? Mandatory seatbelts, drug prohibitions, and restrictions on gambling often rely partly on paternalistic reasoning. Here many liberals become skeptical, arguing that competent adults should generally be free to make poor choices.
-
Prevention of indirect social costs. A common argument is that self-regarding conduct eventually imposes costs on others through families, health systems, insurance pools, crime, or lost productivity. The challenge is that almost every action has indirect social effects. If this principle is allowed unlimited scope, liberty can shrink dramatically.
-
Enforcement of morality. Some argue that conduct may be criminalized because it is morally wrong even if it harms no one. Others reject this entirely. Much of the modern debate over sexuality, gambling, drugs, and speech revolves around this question.
If I were constructing a framework from the goals you listed, I would probably formulate a series of tests:
-
Is there a clearly identifiable harm to non-consenting others?
-
If not, is there a substantial and demonstrable risk of such harm?
-
If not, does the conduct threaten institutions essential to liberty, safety, or prosperity?
-
If not, are the affected persons incapable of informed consent?
-
If none of the above apply, is criminalization likely to reduce overall well-being more than it reduces liberty?
-
Is a less restrictive remedy available, such as civil liability, regulation, taxation, licensing, education, or social pressure?
That last question is important. Criminal law is society's most coercive domestic tool. Arrest, prosecution, fines, imprisonment, and criminal records carry heavy costs. One could argue that conduct should not be criminalized merely because it is undesirable; it should be criminalized only when less severe measures are inadequate.
The result is a kind of presumption of liberty: criminalization requires justification, and the burden of proof lies with those who would prohibit conduct rather than those who would permit it.
The interesting test cases are things like recreational drugs, prostitution, gambling, assisted suicide, helmet laws, and compulsory vaccination. Those are where different philosophies diverge because the first four questions don't produce obvious answers. A framework becomes useful precisely when it helps explain why one reaches a particular conclusion in those difficult cases.
You are right to call out the memory point. I phrased that too casually. A courtroom appearance in which one's livelihood is discussed, one's conduct is publicly judged, and a judge delivers a serious warning is precisely the sort of emotionally charged event that people tend to remember for a very long time. It would not need any special explanation to account for its staying power.
Your second point is more philosophically interesting: the absence of justice from the list of purposes.
I listed the common categories used in criminal justice theory—retribution, deterrence, incapacitation, rehabilitation, and expression—but that does not mean "justice" is absent. Rather, many theorists treat justice as the larger concept within which those goals are evaluated. The omission is revealing, though, because it raises the question: what do we mean by justice?
There are at least two very different ideas:
-
Justice as deserved response
The person who did wrong deserves some punishment because a wrong was committed. This is the retributive idea. It is backward-looking: What does the offender deserve because of what happened? -
Justice as restoring a proper order
The response should repair, as much as possible, the harm, restore respect for the rules, and put society back into balance. This is more forward-looking.
Your Ontario example seems to fit the second idea better. The judge did not ignore the violation; the driver was convicted. The wrong was formally recognized. But the response was designed around future behavior and proportionality rather than simply making the driver suffer.
The interesting question is whether justice is actually a separate goal from the ones I listed or the principle that tells us how to balance them. For example:
-
Deterrence without justice can become excessive punishment ("make an example of him").
-
Rehabilitation without justice can become paternalistic ("we know what is best for you").
-
Public safety without justice can become authoritarian ("anything is acceptable if it reduces risk").
Justice acts as a constraint: the state may pursue good outcomes, but not by any means it chooses.
Your earlier argument about penalties varying fits neatly here. A just system does not merely ask, "Did the person violate a rule?" It asks:
-
How serious was the wrong?
-
Was harm caused?
-
Was it intentional or careless?
-
What response is proportionate?
-
What response serves society without destroying the person unnecessarily?
The judge's decision seems to have reflected that kind of proportional justice. The driver was held accountable, but the court recognized that removing his livelihood would be a disproportionate response under the circumstances.
There is also a connection to your original question about when a state should criminalize. A state that claims to act for the well-being of its citizens still needs justice as a limiting principle. Otherwise, a government could argue that almost any restriction is justified if it claims some benefit.
So perhaps the structure becomes not just:
-
liberty,
-
safety,
-
prosperity,
-
happiness,
but:
-
justice determines when and how the state may sacrifice one of these for another.
That may actually be the missing organizing principle. A benevolent state without justice risks becoming paternalistic; a just state without benevolence risks becoming merely procedural. The challenge is keeping the two in balance.
You are identifying one of the most difficult tensions in modern criminal justice: the difference between justice as a response to a wrong and justice as a vehicle for the emotional needs of those affected by the wrong.
Victim impact statements (including statements by family members of murder victims) are now permitted in many jurisdictions, but they remain controversial for exactly the reason you give. The concern is that they may shift the sentencing decision from what punishment is proportionate for this offender and this offense toward how much suffering the offender has caused, and how much more suffering should be imposed in response.
That is where the danger of "revenge in fancy wrapping" appears. The family’s grief is real and profound, but grief is not necessarily a good measure of what the state should do. Two murders of equal legal and moral seriousness do not become different crimes because one victim leaves behind ten grieving relatives and another leaves behind none.
A justice-based framework might say:
-
The victim's harm is relevant because it helps establish the seriousness of the offense.
-
The victim's circumstances may matter in understanding the crime.
-
The suffering of the family may be acknowledged.
-
But the family's desire for a particular punishment should not determine the sentence.
This distinction is similar to the one you made earlier about the cab driver's livelihood. A person's circumstances matter, but they should matter through principles of fairness and proportionality, not simply because one person's emotional situation is more compelling than another's.
The strongest argument for allowing family testimony is not that it should increase punishment. It is that the criminal trial is one of the few formal settings where victims and families are recognized as human beings rather than merely evidence in a case. A purely state-versus-defendant model can make victims feel invisible.
But that benefit is different from the question of sentencing. One could imagine a system where families are given a voice and recognition, while the judge is instructed that the purpose is understanding the harm, not measuring vengeance.
Your position has a long philosophical lineage: the state takes over the role of punishing precisely so that punishment is not left to private revenge. The argument for public justice is that society replaces "you hurt me, so I hurt you back" with a measured response based on rules and proportionality.
The hard question is where recognition ends and influence begins. A family member saying, "This is how the crime affected our lives" seems different from saying, "Therefore the offender deserves the maximum possible sentence." The first informs the court; the second risks transferring the power of punishment from law to emotion.
So I think your framework would likely distinguish between acknowledgment of harm and determination of penalty. The former can serve justice; the latter belongs to the impartial application of law. That distinction is precisely where many debates over victim impact statements have centered.
That is a very thoughtful skepticism, and it points to a genuine ambiguity in the word justice. It is one of those words that carries so much moral weight that people often use it without agreeing on what they mean.
Your two concerns correspond to two major historical meanings:
-
Justice as retribution — the idea that wrongdoing creates a moral imbalance that must be corrected by imposing a deserved penalty. This is the "cosmic scales" notion: a wrong has occurred, and punishment restores the balance.
-
Justice as revenge with official clothing — the concern that society sometimes takes the understandable anger of victims and gives it the authority of the state.
Both can become problematic. The first can lead to the idea that suffering is inherently valuable because someone "deserves" it. The second can make punishment a socially acceptable form of vengeance.
An alternative is to treat justice less as a mystical property and more as a set of practical constraints on legitimate state action:
-
Fairness: like cases should be treated alike.
-
Proportionality: the response should fit the seriousness of the conduct.
-
Due process: decisions should be made through reliable procedures rather than emotion or power.
-
Impartiality: the outcome should not depend on who the victim or offender happens to be.
-
Respect for persons: even those who commit terrible acts retain some moral status.
Under that approach, justice is not something that must be "balanced" in the universe. It is a discipline for preventing arbitrary or excessive uses of power.
This also fits with your earlier discussion of paternalism. A state might sincerely want to improve citizens' welfare, but without a concept like justice it could justify almost anything: "This will make people safer, healthier, happier, or more prosperous." Justice provides the boundaries.
I suspect your discomfort is partly with the word's emotional baggage. Terms like "fairness," "proportionality," and "legitimate authority" may describe what you are trying to capture more precisely. They are less likely to suggest either vengeance or metaphysical accounting.
One interesting observation is that many legal systems have gradually moved in that direction. Older concepts of punishment often emphasized retribution—"an eye for an eye" in its harsher interpretations. Modern criminal law tends to justify punishment more through public purposes: deterrence, rehabilitation, incapacitation, and maintaining a lawful order. Yet the language of justice remains because people still want a term that says the state is not merely calculating outcomes; it is also obligated to treat people properly.
So perhaps a useful formulation for your framework would be:
The state should not seek justice as revenge or cosmic compensation. It should seek justice as the fair and proportionate exercise of legitimate power.
That definition keeps the useful part of the word while avoiding the meanings you find troubling.