Putting pen to paper on the world’s big issues 

Three Melbourne Grammar students have had their essays shortlisted in the international John Locke Institute Global Essay competition.

Jerry Pei (Year 12) has been shortlisted with a science and technology essay, Zihe Wang (Year 12) with a law essay, Andrew Yuan (Year 10) with an economics essay.

The John Locke Institute “encourages young people to cultivate the characteristics that turn good students into great writers: independent thought, depth of knowledge, clear reasoning, critical analysis, and persuasive style”.

Essays are judged by a panel of senior academics drawn from universities including Oxford, Cambridge, Harvard, Princeton, and Stanford.

The below excerpts from their essays have been edited.

Should humans be polite to Chat GPT?

Science and Technology
Jerry Pei (Year 12)

Long before AI, Aristotle’s Nicomachean Ethics offered a useful framework for understanding why everyday interactions matter morally. [18] Aristotle’s theory of habituation argues that character is not mainly formed through innate goodness, but through repeated action. We become the kind of people we are by repeatedly undertaking certain actions. As Aristotle suggests, virtue is “made perfect by habit,” meaning that moral character is gradually built through consistent practice rather than occasional intention.

This is supported by modern research, which suggests that repeated behaviours in similar contexts gradually become more automatic, while habits can also be understood as learned responses strengthened through repetition in recurring situations.[19][20] As AI interaction becomes increasingly common, it is becoming one of these recurring situations. Many people interact with AI tools daily, meaning that the tone users utilize in these interactions is not an isolated act. It becomes part of a repeated communicative pattern, and consequently shapes our character within real-world social interactions. Indeed, studies demonstrate that when people perceive AI as more humanlike or conscious, the way they treat AI influences how they treat other people, because similar social-cognitive processes are activated.[21] Thus, the carry-over effect is not merely hypothetical, but has been empirically shown to influence human conversations. Consequently, even though AI systems lack human emotions, users should still maintain conversational politeness, as the moral significance lies less in what AI experiences or deserves morally and more in what these repeated interactions train humans to become.

In essence, the question of whether humans should be polite to AI systems is not about the moral status of machines but about the effects of interaction on human behaviour and communication. The evidence suggests that politeness is not simply an optimisation strategy, as its effects on output vary. However, it can shape interactions that are more pleasant, cooperative, and useful for the user. More importantly, repeated interaction styles can influence users’ own communicative habits, reinforcing either cooperative or dismissive patterns within our day-today interactions with others. For these reasons, ordinary conversational politeness toward AI systems such as ChatGPT is practical and beneficial. It supports clearer and more cooperative exchanges, allowing users to learn more from these interactions while preserving social norms that extend beyond human–AI communication. The aim is not exaggerated deference, but a baseline level of natural politeness that reflects how we would speak in ordinary human conversation.

To what extent should criminal sentencing take into account the effect on the perpetrator’s family?

Law
Zihe Wang (Year 12)

At its core, criminal punishment is justified only insofar as it is directed at the culpable offender. This reflects the foundational principle that criminal liability, and therefore punishment, must track personal responsibility. However, sentencing frequently harms individuals who bear no such responsibility. The imprisonment of an offender may deprive a family of its primary source of income, subject relatives to social stigma, and cause profound emotional disruption, particularly for dependent children. As demonstrated in a large-scale analysis, children of incarcerated parents face significantly higher risks of psychological harm, educational disruption, and later offending.[2] If sentencing is applied without regard to these effects, it undermines the principle that punishment should be proportionate to individual culpability, extending its impact beyond those who deserve it. This principle is clearest in cases involving dependent children, who are entitled to heightened protection from the state.

Beyond avoiding unjust harm to innocent family members, sentencing should also take family impact into account where doing so meaningfully advances rehabilitation and, by extension, broader societal stability. Unlike proportionality, the principle that punishment should correspond to the seriousness of the offence and the offender’s culpability, rehabilitation is forward-looking, concerned with the offender’s capacity to reintegrate into society and avoid future offending. For instance, an offender with stable family support may be better able to secure employment, maintain lawful routines, and avoid reoffending than one whose family ties have been unnecessarily disrupted. In this context, preserving family relationships is critical. Stable familial support structures have been consistently recognised as factors that reduce recidivism, as demonstrated by research finding that inmates who received visits were less likely to reoffend.[11]

Furthermore, such structure assists with post-release reintegration, and provides both practical and emotional frameworks through which offenders can re-establish lawful lives.[12] Unnecessary disruption of these relationships may therefore weaken rehabilitation and increase future social costs. This does not imply that punishment should be reduced simply because an offender has a supportive family; rather, it supports the careful structuring of sentences, such as through non-custodial alternatives, suspended sentences (where imprisonment is imposed but not immediately served unless conditions are breached) or timing considerations, where doing so preserves rehabilitative pathways without compromising the seriousness of the offence. A sentencing system that disregards these effects risks inflicting avoidable hardship on innocent parties, and weakens its own capacity to promote the social order it was designed to uphold.

_{[2] Murray, J., Farrington, D. P., & Sekol, I. (2012). “Children’s antisocial behavior, mental health, drug use, and educational performance after parental incarceration: A systematic review and meta-analysis.” Psychological Bulletin, 138(2), 175–210.}

_{[11] Bales, W. D., & Mears, D. P. (2008). “Inmate social ties and the transition to society: Does visitation reduce recidivism?” Journal of Research in Crime and Delinquency, 45(3), 287–321.}

_{[12] Visher, C. A., & Travis, J. (2003). “Transitions from prison to community: Understanding individual pathways.” Annual Review of Sociology, 29, 89–113.}

Technology now allows personalised pricing. If this came to be widely used, what effects should we expect?

Economics
Andrew Yuan (Year 10)

Nowadays, determining a product’s worth involves individual valuations more than ever. Firms can access increasingly rich data on consumers’ location, device type and online history and thereby estimate each consumer’s maximum willingness to pay (WTP), approximating first-degree price discrimination.[1] [2] About 21% of UK firms already possess the necessary infrastructure, with a further 33% planning to.[3]

This essay argues that widely employed personalised pricing (PP) would change the consumer segments protected by competition, often harming the already disadvantaged.

In practice, consumers still purchase goods but derive much less economic value from doing so. Such a transaction, even if economically efficient, is not necessarily socially beneficial if nearly all the gains are captured by firms.

In a field experiment on ML PP, Dubé and Misra (2023) found that PP increased firm profits by 19% relative to an optimised uniform price, while total CS fell by 23%.[8] This is worrying because ownership of large firms and financial assets is highly concentrated, with the top 1% holding 50.2% of corporate equities and mutual fund shares in the USA in Q3 2025.[9] While some of this is held indirectly through pensions and mutual funds, such holdings are themselves unequally distributed because richer households hold larger financial portfolios. Consequently, the gains in producer surplus accrue disproportionately to wealthier groups rather than being broadly distributed. The result is less total utility-weighted welfare due to the diminishing marginal utility wealthier people experience.

Furthermore, although 60% of consumers may receive lower prices, PP can still worsen welfare if higher prices are concentrated on constrained consumers with inelastic demand. It is the case that poverty almost always reduces substitutability for goods and services.[10] Consumers with less money, transport access and time cannot buy in bulk, travel to cheaper suppliers or absorb the costs of searching for alternatives. As a result, 24% of British households already experience a ‘poverty premium’, costing low-income households an annual average of 430 pounds.[11] This lack of choice for socioeconomically disadvantaged consumers is easily exploitable by PP algorithms. Moreover, lower-income households generally spend a larger proportion of their income on necessities which have very inelastic demand.[12] Thus, the welfare loss that occurs because high WTP reflects how constrained a consumer is and that constraint is both more common and more damaging among poorer consumers.

_{[1] FTC Surveillance Pricing Study Indicates Wide Range of Personal Data Used to Set Individualized Consumer Prices. (2025, January 17). Federal Trade Commission. Retrieved June 18, 2026, from https://www.ftc.gov/news-events/news/press-releases/2025/01/ftc-surveillance-pricing-study-indicates-wide-range-personal-data-used-set-individualized-consumer}

_{[2] OECD. (2018). Personalised pricing in the digital era: Background note by the Secretariat (DAF/COMP(2018)13). https://one.oecd.org/document/DAF/COMP(2018)13/en/pdf}

_{[3] Lombardelli, C., & Patel, R. (2026, April 7). This time it’s personal: The rise of dynamic, personalised pricing and what it means for inflation. Bank of England. https://www.bankofengland.co.uk/bank-insights/2026/this-time-its-personal-the-rise-of-dynamic-personalised-pricing-and-what-it-means-for-inflation}

_{[8] Dubé, J.-P., & Misra, S. (2017). Scalable price targeting (NBER Working Paper No. 23775). National Bureau of Economic Research. https://www.nber.org/system/files/working_papers/w23775/w23775.pdf}

_{[9]Board of Governors of the Federal Reserve System. (2026, March 27). Distribution of household wealth in the U.S. since 1989. https://www.federalreserve.gov/releases/z1/dataviz/dfa/distribute/table/}

_{[10]Goodman, C. S. (1968). Do the poor pay more?. Journal of Marketing, 32(1), 18-24. https://doi.org/10.1177/002224296803200105.}

_{[11]Fair By Design. (2023, February 21). The local poverty premium. https://fairbydesign.com/policy-advocacy/the-local-poverty-premium/}

_{[12]Office for National Statistics. (2022, July 18). Family spending in the UK: April 2020 to March 2021. https://www.ons.gov.uk/peoplepopulationandcommunity/personalandhouseholdfinances/expenditure/bulletins/familyspendingintheuk/april2020tomarch2021}