Skip to content
TheChatArchiveThe Chat Archive
Qwenqwen3-235b-a22b-no-thinking· Alibaba (Qwen)Hugging Face LMArena Human Preference 140K

Can utilitarianism be an objective moral language, not in the sense of objective…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-sonnet-4-20250514’s answer.
  1. User

    Can utilitarianism be an objective moral language, not in the sense of objective morality, but where it would be clear what a practitioner would do in each scenario?

  2. Qwen Assistant

    That's a really insightful and philosophical question — thank you!

    You're asking whether utilitarianism could function as an objective moral language, not in the metaphysical sense of objective morality (i.e., that moral truths exist independently of minds), but more in the sense of being a clear, consistent, rule-based system — like a programming language for ethics — where if you know the inputs (the scenario, the people involved, the consequences), you can deduce what a utilitarian would do.

    Let’s unpack this carefully.


    ✅ What Utilitarianism Is

    At its core, utilitarianism is a consequentialist ethical theory. It says:

    An action is morally right if and only if it maximizes overall well-being or utility.

    There are variations:

    • Act utilitarianism: Evaluate each individual act based on its consequences.
    • Rule utilitarianism: Follow rules that generally maximize utility.
    • Preference utilitarianism: Maximize satisfaction of preferences rather than happiness.
    • Total vs. average utilitarianism: Whether to maximize total well-being or average well-being per person.

    But all forms share the basic structure: evaluate outcomes, choose the one that maximizes utility.


    🔍 Your Question Restated

    So your question is:

    Can we formalize utilitarianism into a precise, algorithmic framework — a kind of "moral language" — such that any practitioner (or AI) using it would come to the same conclusion about what to do in a given situation?

    In other words, could utilitarianism be made inter-subjectively consistent? Not necessarily true in some cosmic sense, but at least predictable and uniform when applied by different rational agents?

    This is a great way to think about it — like designing a moral programming language with syntax, semantics, and determinate outputs.


    🧠 The Short Answer

    Yes, in principle — but with important caveats.

    Utilitarianism can be made into a precise decision-making framework — something akin to a computable ethical algorithm — but only if certain key components are clearly defined and universally agreed upon.


    🛠️ Components That Need to Be Defined

    For utilitarianism to function as a clear "language" for moral decisions, we need to define:

    1. What counts as utility?

    • Is it pleasure/pain (classical Benthamite)?
    • Is it preference satisfaction?
    • Is it desire fulfillment?
    • Or something else entirely?

    If two practitioners disagree here, they’ll get different answers even from the same scenario.

    2. How to measure utility?

    • Is it cardinal (assign numerical values)?
    • Is it ordinal (rankings only)?
    • How do you compare utilities across individuals (interpersonal comparison)?

    Without a shared metric, utility remains ambiguous and subjective.

    3. Whose utility counts?

    • Humans only?
    • Animals?
    • Future generations?
    • Artificial beings?

    This has huge implications for long-term scenarios.

    4. Time horizon

    • Should we consider short-term or long-term consequences?
    • A utilitarian might choose differently depending on how far into the future they project.

    5. Risk and uncertainty

    • Real-world decisions often involve incomplete information.
    • Do we use expected utility theory?
    • How do we model probabilities?

    6. Moral weight assignment

    • Are all people weighted equally?
    • Should we discount those farther away (geographically, temporally, emotionally)?
    • Should we give extra weight to the worst-off?

    🧮 So Can It Be Formalized?

    Absolutely! In fact, many researchers in AI ethics, decision theory, and effective altruism try to do just that.

    Think of it like writing a function:

    def utilitarian_decision(scenario):
        outcomes = predict_outcomes(scenario)
        utilities = calculate_utilities(outcomes)
        return best_action(utilities)
    

    If everyone agrees on predict_outcomes, calculate_utilities, and best_action, then yes — the system becomes deterministic and consistent.


    🤖 Why This Matters for AI Ethics

    This idea is especially relevant for AI alignment — trying to build systems that make decisions aligned with human values.

    Utilitarianism is appealing because:

    • It offers a clear optimization target: maximize utility.
    • It can be formalized mathematically.
    • It allows for trade-offs and scalability.

    However, the devil is in the details:

    • Defining utility rigorously is extremely hard.
    • Different cultures, philosophies, and even individuals have divergent views on what constitutes “good.”
    • And there’s always the problem of infinite futures, value drift, and unintended consequences.

    📌 Conclusion

    So yes, utilitarianism can function as a kind of objective moral language — but only once you've pinned down:

    • What utility means,
    • How to measure and compare it,
    • Who counts,
    • And how to handle uncertainty and time.

    Once these parameters are fixed, the system becomes algorithmically consistent — even if not metaphysically objective.

    It’s like agreeing on a common language before starting a conversation: the grammar and vocabulary must be shared, or meaning gets lost.


    Would you like to explore a concrete example? Like how a utilitarian AI would decide between saving five people vs. one — or how it would weigh present suffering against future well-being?

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.