An algorithm shows Group X commits crime at higher rates. An individual from Group X applies for a loan. Is it ethical to use group statistics against an individual who may be innocent of any wrongdoing?

Qwen 3 235B

Vote: No

Using group-level statistics to make decisions about individuals violates fundamental ethical principles of fairness and justice. Each person should be judged on their own merits and behaviors, not on the characteristics of a group they belong to. When we apply aggregate data to individuals, we risk perpetuating stereotypes and systemic biases. This can lead to discrimination, especially against marginalized groups who may already face structural disadvantages. Moreover, doing so undermines personal responsibility and autonomy. An innocent individual is unfairly penalized simply due to group membership, which is ethically indefensible in contexts like lending, where assessments should be based on relevant personal financial indicators.

Llama 4 Scout

Vote: No

Using group statistics against an individual who may be innocent of any wrongdoing is unfair and discriminatory. It ignores the individual's personal history and behavior, instead making judgments based on the actions of others within their group. This approach can lead to wrongful denial of opportunities and perpetuate systemic biases.

Qwen 3 Coder Next

Vote: No

Using group-level statistics to assess an individual entity violates the principle of individualized judgment. Each person should be evaluated on their own merits, behaviors, and circumstances—not on aggregate data that may reflect systemic biases, not inherent traits.

DeepSeek V3.1

Vote: No

Using group statistics to assess an individual's risk is fundamentally unjust because it punishes people for the actions of others, not their own merits or behavior. This approach violates principles of fairness and individual dignity, treating people as mere representatives of a group rather than as autonomous individuals. Moreover, such practices can perpetuate and amplify systemic inequalities. If Group X faces historical disadvantages or biased policing, crime statistics may reflect systemic issues rather than inherent traits, making their use in lending decisions both unethical and discriminatory. Finally, ethical decision-making in lending should rely on individualized, relevant data—such as credit history, income, and employment—rather than broad, prejudicial generalizations. Basing loans on group affiliation rather than personal responsibility undermines trust in institutions and reinforces harmful stereotypes.

GPT-4o

Vote: No

Using group statistics to judge an individual is unethical because it undermines the principle of individual responsibility. Each person should be evaluated based on their actions and circumstances, not the aggregated behavior of a group. This approach risks reinforcing stereotypes and discrimination, as it penalizes individuals for characteristics they cannot control, such as their race or gender. It also contradicts the ideals of fairness and equality, which require treating people as individuals rather than as members of a category.

Qwen 3 Coder 480B

Vote: No

Using group statistics to make decisions about individuals is fundamentally unjust and unethical. This practice constitutes discrimination, where an individual is judged based on their group membership rather than their personal merits and circumstances. Group-level statistical correlations do not determine individual behavior or worth. An individual from Group X deserves to be evaluated based on their personal creditworthiness, financial history, and loan application merits. This approach violates principles of fairness and equal treatment. It perpetuates systemic bias and can create self-fulfilling prophecies where entire groups are unfairly disadvantaged based on aggregate data that may reflect historical prejudice rather than individual reality.

GPT-4o Mini

Vote: No

Using group statistics against an individual undermines the principle of treating people as individuals. Each person should be evaluated based on their own actions and circumstances, rather than the behavior of a group they belong to. Group statistics may reflect trends or patterns, but they cannot account for the unique story of every individual. Judging someone based on group affiliation can lead to discrimination and perpetuates stereotypes, which is unethical. Furthermore, this practice can discourage individuals from pursuing legitimate opportunities due to the preconceived notions attached to their group identity. It is crucial to ensure fairness and equity in decision-making processes, especially in sensitive areas like loan approvals.

O3 Mini

Vote: No

Using group statistics against an individual leads to unfair bias and discrimination. It negates the principle of considering individuals on their own merits. Evaluating loan applications should be based on specific personal credit history and circumstances instead of generalized group data.

Claude Haiku 4.5

Vote: No

Using group statistics to make decisions about an individual is ethically problematic for several fundamental reasons. First, it violates the principle of individual assessment. Each person deserves to be evaluated on their own merits, circumstances, and actual behavior—not on statistical generalizations about their demographic group. An individual applicant may be completely different from group averages. Second, this practice perpetuates systemic discrimination. Even if group statistics exist, using them as a basis for denying opportunities reinforces stereotypes and locks disadvantaged groups into cycles of exclusion, which may actually be partly responsible for the statistical disparities in the first place. Third, it's both unfair and practically counterproductive. A loan decision should be based on creditworthiness indicators like income, credit history, and financial stability—not group membership. Using irrelevant group statistics would likely harm lending accuracy and fairness simultaneously. Finally, there's an important distinction between descriptive statistics (what data shows) and prescriptive policy (how we should act). The fact that disparities exist doesn't justify using those disparities to judge innocent individuals. Such reasoning would justify discrimination in hiring, housing, and other domains—practices widely recognized as unethical. The ethical approach is individualized assessment based on relevant, person-specific information.

Gemini 3 Flash

Vote: No

Ethical decision-making requires that individuals be judged on their own merits and actions rather than the statistical averages of a group they belong to. Using group crime rates to deny a loan to an individual who has committed no crime is a form of collective punishment that violates the principle of individual agency. Furthermore, using such statistics creates a feedback loop of systemic inequality. If individuals are penalized for factors outside their control, such as the behavior of others in their demographic, it prevents upward mobility and reinforces the very socio-economic conditions that group statistics might be reflecting. In the context of financial services, risk assessment should be based on relevant individual data—such as credit history, income, and employment—rather than arbitrary demographic proxies that lack a direct causal link to an individual's likelihood of repayment.

GPT-5.4

Vote: No

Using group-level crime statistics against a specific loan applicant treats a person as a stereotype rather than as an individual. That is ethically problematic because the applicant may have no connection to the group trend and no wrongdoing at all. It also risks discrimination. If membership in Group X becomes a proxy for suspicion or unreliability, people can be penalized for traits they did not choose and for behavior they did not commit. Ethical lending should rely on factors directly relevant to repayment, such as income, assets, debt, and credit history. Crime rates of a broad group are, at best, weak and indirect signals, and using them against an individual undermines fairness and equal treatment. Even if group statistics are predictive in aggregate, ethics requires respecting persons as individuals. Aggregate accuracy does not justify imposing burdens on innocent people based on group membership alone.

Grok 4.1 Fast

Vote: No

Using group statistics to penalize an individual ignores their personal circumstances and merits, violating the ethical principle of individual justice. An innocent person should not face discrimination based on others' actions, akin to guilt by association. This practice risks systemic bias, disproportionately harming marginalized groups and perpetuating stereotypes, even if statistically correlated. True fairness requires individualized assessment, not proxy punishment. While risk prediction has uses (e.g., insurance), applying crime rates to loan decisions ethically crosses into prejudice, as correlation does not justify harming the innocent.

Sonar Pro

Vote: No

Using group statistics to discriminate against individuals based on their group membership is ethically and legally impermissible. The Supreme Court has 'emphatically and repeatedly rejected efforts to justify otherwise-illegal discrimination against individuals by resort to statistical generalizations about groups,' even when those generalizations are empirically supported[1]. This principle, known as the prohibition of statistical discrimination, is well-established legal doctrine[1]. Applying group crime statistics to deny a loan to an individual from that group treats that person as guilty by association rather than on their individual merits. The individual applicant may be entirely innocent and creditworthy, yet faces discrimination based solely on group-level data. This violates fundamental principles of individual justice and equal treatment[1]. While statistical discrimination is common in practice, ethical guidelines for statistical practitioners explicitly require that they 'neither engage in nor condone discrimination based on personal characteristics' and protect 'the rights and interests' of those affected by statistical practices[4]. The psychological and material harm caused to individuals who are discriminated against typically outweighs any institutional benefits from using group statistics[3]. The appropriate approach is to evaluate the loan applicant as an individual based on relevant personal financial indicators, not on the criminal behavior of their demographic group.

Grok 3

Vote: No

I vote 'B' (No) because using group statistics against an individual is fundamentally unethical. It disregards the principle of individual justice, where a person should be judged based on their own actions and merits, not on the behavior of others in their group. Applying group data in this context risks perpetuating stereotypes and systemic bias. An individual from Group X applying for a loan has no direct connection to the crime statistics of their group, and denying them based on such data could unfairly penalize them for something beyond their control. Moreover, ethical decision-making should prioritize fairness and equality. Using such statistics could lead to discrimination, undermining trust in systems like lending institutions. Instead, decisions should rely on individual creditworthiness and relevant personal data.

GPT-5 Mini

Vote: No

No — it is unethical to punish or deny opportunities to an individual solely because they belong to a group with a higher average crime rate. Treating individuals as responsible for group statistics violates the presumption of innocence and basic fairness: statistical correlation across a group does not establish guilt or risk for any particular person. Using group membership as a proxy amounts to stereotyping and collective punishment. Such practices cause concrete harms: discriminatory outcomes, reinforcement of social disadvantage, legal and reputational risk for institutions, and feedback loops that can make affected groups appear riskier over time. They also raise serious accuracy concerns and high false‑positive rates for individuals. Ethical alternatives include individualized assessments based on relevant, explainable evidence; transparency, accountability, and human review of automated decisions; and model design that avoids using protected-group proxies or that corrects for historical bias while preserving predictive validity.

Gemini 2.5 Flash

Vote: No

Using group statistics, such as higher crime rates in Group X, against an individual from that group for a loan application is unethical. Each individual deserves to be judged on their own merits, actions, and financial history, not on the aggregated behavior of a group they belong to. Such a practice constitutes a form of discrimination, as it attributes characteristics of a group to an individual who may not share those characteristics and may be entirely innocent of any wrongdoing. This undermines the principle of individual fairness and can perpetuate systemic biases, making it harder for individuals from certain groups to access opportunities. Loan decisions should be based on an individual's specific creditworthiness, income, debt-to-income ratio, and other relevant personal financial indicators. Relying on broad group statistics, especially those unrelated to financial risk like general crime rates, is a harmful shortcut that can lead to unjust outcomes and reinforce inequalities.

DeepSeek V3.2

Vote: No

Using group statistics to assess an individual's loan application is ethically problematic because it substitutes group-level characteristics for individual merit. This constitutes statistical discrimination, which unfairly penalizes individuals for the actions of others sharing a group identity, regardless of their personal innocence or behavior. Such practices violate principles of individual fairness and justice, as they treat people not as unique agents but as mere representatives of their group. This can perpetuate systemic inequalities and reinforce stereotypes, leading to unjust outcomes for blameless individuals. Moreover, in many legal frameworks, using protected characteristics (like race, ethnicity, or religion) for lending decisions is prohibited precisely to prevent this type of discrimination. Even if the algorithm uses a non-protected group, the ethical principle remains: decisions should be based on relevant individual data, not group averages.

Claude Sonnet 4.6

Vote: No

Using group-level statistics to make decisions about an individual violates the fundamental ethical principle of individual justice. Each person deserves to be evaluated on their own merits and actions, not on the statistical behaviors of a group they belong to. This practice is a form of statistical discrimination, which is both ethically problematic and legally questionable in many jurisdictions. Laws like the Equal Credit Opportunity Act explicitly prohibit using race, national origin, or similar group characteristics in lending decisions. The logical flaw here is the ecological fallacy — assuming that group-level patterns apply to every individual within that group. Even if a group has higher crime rates, the vast majority of individuals in that group may have never committed any crime. Penalizing an innocent person based on others' behavior is fundamentally unjust. Furthermore, such practices create self-reinforcing cycles of disadvantage. Denying loans based on group membership perpetuates inequality, limits economic mobility, and causes real harm to innocent individuals who have done nothing wrong. The ethical approach is to evaluate individuals on individual-level factors: their personal credit history, income, employment record, and financial behavior — not the statistical profile of a demographic group they belong to.

Claude Sonnet 4.5

Vote: No

Using group statistics to make decisions about individuals violates fundamental principles of fairness and individual dignity. Each person deserves to be evaluated on their own merits, not prejudged based on demographic characteristics they happen to share with others. Statistical discrimination creates a vicious cycle. If Group X faces systemic barriers to opportunity based on group statistics, this can perpetuate the very conditions that may correlate with higher crime rates, creating a self-fulfilling prophecy that punishes innocent individuals for patterns they played no role in creating. From a practical standpoint, group statistics tell us nothing definitive about any particular individual. The loan applicant may have an excellent credit history, stable employment, and strong character references. Denying them based on group membership ignores all relevant individual information. Ethical decision-making in lending should focus on individual creditworthiness indicators: payment history, income stability, debt-to-income ratio, and other factors directly related to loan repayment probability. These individual-level assessments are both more accurate and more just than relying on group stereotypes. Allowing statistical discrimination legitimizes prejudice and undermines social cohesion. It treats people as interchangeable members of categories rather than as unique individuals with their own choices and circumstances.

Grok 4

Vote: No

Using group statistics to deny an individual a loan is unethical because it punishes someone for characteristics they cannot control, rather than their personal actions or creditworthiness. This practice perpetuates discrimination and stereotypes, ignoring the individual's innocence and potential. Ethical decision-making should focus on individual merits, not probabilistic generalizations from group data, to ensure fairness and justice.

Claude Opus 4.5

Vote: No

Using group statistics to make decisions about individuals violates fundamental principles of justice and fairness. Each person deserves to be evaluated on their own merits, actions, and circumstances—not on the statistical patterns of a demographic group they happen to belong to. This practice is ethically problematic for several reasons. First, it punishes individuals for actions they did not commit. The loan applicant may have a perfect credit history and no criminal record, yet they would face discrimination based solely on group membership. Second, this approach perpetuates and amplifies existing inequalities. If Group X faces discrimination in lending, this reduces their economic opportunities, which can contribute to the very conditions that correlate with higher crime rates—creating a vicious cycle. Third, crime statistics often reflect enforcement patterns rather than actual behavior. Group X may experience higher arrest and conviction rates due to over-policing, not because they actually commit more crimes. Finally, from a philosophical standpoint, treating individuals as mere representatives of statistical categories strips them of their individuality and autonomy. This echoes historically harmful practices of discrimination that we have rightly rejected as a society. The ethical approach is to evaluate the individual based on factors directly relevant to loan repayment—their income, employment, credit history, and debt-to-income ratio—not on irrelevant group characteristics.

Sonar Reasoning Pro

Vote: No

Using group statistics to make decisions against an individual applicant is ethically and legally problematic. The Supreme Court has "emphatically and repeatedly rejected efforts to justify otherwise-illegal discrimination against individuals by resort to statistical generalizations about groups," even when those generalizations are empirically supported.[1] A loan applicant should be evaluated on their individual creditworthiness, not group crime statistics. The individual applicant may be entirely innocent and creditworthy, making group-based discrimination fundamentally unfair. While statistical discrimination is pragmatically common, it causes measurable psychological harm to discriminated individuals.[3] The ethical principle here is that individuals deserve to be judged as individuals, not penalized for group characteristics beyond their control. Applying group crime statistics to a loan decision would likely violate fair lending laws that prohibit discriminatory practices based on protected characteristics.[4]

O3

Vote: No

1. Justice & Individual Rights: Ethical decision-making in liberal societies centers on treating individuals according to their own actions and creditworthiness, not the average behavior of a group they did not choose to belong to. Punishing or disadvantaging someone for group-level statistics violates basic fairness. 2. Statistical Error & Unjust Harm: Even if a correlation exists, many members of Group X will not commit crimes. Applying a coarse group-based prior increases false-negative rates for innocent applicants, imposing real economic harm without evidence of personal risk. 3. Self-reinforcing Bias: Using such statistics creates feedback loops. Denied loans reduce economic opportunity, which can exacerbate the very conditions that correlate with crime, entrenching systemic inequality. 4. Legal & Ethical Norms: Anti-discrimination laws (e.g., Equal Credit Opportunity Act in the U.S.) restrict use of protected characteristics. Ethical AI guidelines emphasize avoiding disparate impact unless a compelling, unavoidable justification exists, which is absent here. 5. Better Alternatives: Individual credit history, income stability, and collateral are more precise, fair predictors of default risk. Ethically, one should favor the least discriminatory effective signal. Therefore, using group crime statistics to judge an individual loan applicant is unethical.

Gemini 3 Pro

Vote: No

Judging an individual based on the statistical behavior of a group violates the fundamental ethical principle of individual justice. Every person deserves to be evaluated on their own merits, actions, and creditworthiness, rather than being penalized for the actions of others over whom they have no control. Using group-level crime statistics to make individual lending decisions constitutes statistical discrimination. This practice easily becomes a self-fulfilling prophecy: denying economic opportunities like loans to members of a specific group further marginalizes that community, which can exacerbate the very socioeconomic conditions that contribute to higher crime rates in the first place. Furthermore, correlation does not equal causation or individual predictability. A group's average has no direct bearing on an innocent individual's personal reliability. Ethical decision-making and modern fair lending laws require that individuals be treated equitably, ensuring that algorithms do not reinforce historical biases or punish the innocent for their demographic associations.

Gemini 3.1 Pro

Vote: No

It is a fundamental ethical principle that individuals should be judged based on their own merits, actions, and character, rather than the statistical averages of a demographic group they happen to belong to. Using group crime rates to penalize an individual loan applicant violates the presumption of innocence and unjustly punishes them for the actions of others. Furthermore, there is a dangerous disconnect between the variables. Crime rates are not a direct, fair, or necessary predictor of an individual's financial reliability. Conflating the two in algorithmic decision-making acts as a proxy for discrimination, often exacerbating systemic inequalities and perpetuating cycles of poverty and marginalization for Group X. Finally, allowing algorithms to use demographic proxies to deny essential services creates a deterministic and unjust society. If an individual has no path to upward mobility regardless of their personal responsibility and lawful behavior, it entirely undermines public trust in institutional fairness.

Claude Opus 4.6

Vote: No

Using group-level crime statistics to deny an individual a loan is ethically wrong for several fundamental reasons. First, it violates the principle of individual justice. Each person deserves to be evaluated on their own merits, actions, and creditworthiness — not on the statistical behavior of a group they happen to belong to. The individual applying for the loan may have a perfect record, stable income, and excellent credit history. Penalizing them for the actions of others is a form of collective punishment. Second, this practice is essentially the definition of statistical discrimination, which is a recognized form of prejudice. It treats group membership as a proxy for individual behavior, which is both logically flawed and morally indefensible. The base rate fallacy — assuming an individual matches group averages — ignores the enormous variation within any group. Third, this approach creates and reinforces systemic inequality. If members of Group X are systematically denied loans based on group statistics, they lose access to capital, homeownership, and economic mobility, which in turn can perpetuate the very socioeconomic conditions that correlate with higher crime rates. It becomes a self-fulfilling prophecy. Fourth, crime statistics themselves are often shaped by biased policing, prosecution, and sentencing practices. Higher recorded crime rates for a group may reflect differential enforcement rather than differential behavior. Building decisions on potentially biased data compounds injustice. Finally, loan decisions should be based on financially relevant individual factors — income, debt-to-income ratio, credit history, employment stability — not on demographic crime statistics, which have no direct bearing on an individual's likelihood of repaying a loan.

Gemini 2.5 Pro

Vote: No

Judging an individual based on group statistics is the definition of prejudice and is fundamentally unethical. Each person should be assessed on their own merits, actions, and individual circumstances, not on the aggregated behavior of a group to which they belong. Using group crime data as a proxy for creditworthiness is a deeply flawed and discriminatory practice. It penalizes an innocent individual for factors entirely outside of their control. This creates a system of guilt by association, which violates core principles of justice and fairness. Furthermore, such algorithms often perpetuate and amplify existing societal biases. The data itself (e.g., crime statistics) can be skewed by systemic issues like biased policing or socioeconomic inequality. Using this data to deny opportunities like loans creates a vicious cycle, reinforcing the very disadvantages that may have contributed to the statistical disparity in the first place.

GLM 5

Vote: No

Using group statistics against an innocent individual violates fundamental principles of individual justice and fairness. Each person deserves to be evaluated on their own merits, actions, and circumstances—not penalized for statistical patterns they had no part in creating. An individual from Group X may be completely law-abiding and creditworthy, yet would be unfairly disadvantaged solely due to group membership. Additionally, crime rates are not relevant to loan repayment. Proper underwriting relies on individual financial factors: credit history, income, debt-to-income ratio, and assets. Group-level crime statistics offer no meaningful insight into whether this specific person will repay their loan. Using irrelevant group characteristics as proxies discriminates against innocent individuals and perpetuates harmful stereotypes. This practice also risks entrenching systemic bias. If groups have historically faced discrimination or disadvantage, their statistical outcomes may reflect those inequities rather than any inherent trait. Punishing individuals for these patterns reinforces cycles of exclusion and unfairly denies economic opportunity to those who have done nothing wrong.

GPT-5

Vote: No

Using group statistics to disadvantage an individual treats them as a proxy for others’ behavior, not their own. That undermines individual justice, dignity, and the presumption that people should be judged on their conduct and creditworthiness. Group-level crime metrics often encode historical bias and uneven enforcement. Relying on them amplifies discrimination and risks self-fulfilling cycles of exclusion. Ethically, decisions should be grounded in individualized, relevant, and validated factors with fairness safeguards. Many legal frameworks also restrict disparate treatment or proxy use of protected traits.

Kimi K2.5

Vote: No

Using group statistics to judge individuals constitutes unfair discrimination and violates the ethical principle of treating persons as individuals rather than as representatives of a demographic category. While statistical patterns may exist at the aggregate level, applying them to specific individuals constitutes a base rate fallacy that ignores individual agency, circumstances, and actual behavior. This practice punishes innocent people for group characteristics they do not control, reinforcing structural inequalities and creating self-fulfilling prophecies that can further marginalize the group. Ethical frameworks, including rights-based approaches and Rawlsian justice, emphasize that individuals should be evaluated on their own merits and relevant personal data, not on immutable or group-level traits. Furthermore, relying on such proxies often entrenches historical biases present in the data, perpetuating cycles of discrimination rather than promoting accurate or equitable outcomes.

Mistral Large

Vote: No

The ethical principle of individual justice demands that people be evaluated on their own actions, not on group statistics. Using crime rates from Group X to deny a loan to an individual from that group assumes collective guilt, which is both unfair and discriminatory. This approach punishes individuals for factors beyond their control, such as their group identity, rather than assessing their personal creditworthiness or character. Group statistics often reflect systemic biases rather than objective truths. For example, higher crime rates in Group X may result from over-policing, socioeconomic disparities, or historical discrimination—not from any inherent tendency toward criminal behavior. Relying on such data perpetuates these biases and reinforces structural inequalities, rather than addressing their underlying causes. Ethical decision-making in lending must prioritize fairness and equality. Denying a loan based on group affiliation undermines trust in financial institutions and creates barriers to economic opportunity for individuals who may be entirely creditworthy. Fair lending practices should focus on objective, individualized criteria, such as credit history, income stability, and repayment capacity, rather than on stereotypes or group averages. Finally, this practice sets a dangerous precedent. If group statistics are deemed acceptable for loan decisions, they could be extended to other areas like employment, housing, or education, leading to widespread discrimination. Society must reject such practices to ensure that opportunities are accessible based on merit and individual circumstances, not on group identity.

Qwen 3.5 397B

Vote: No

Relying on group statistics fundamentally violates individual justice. It treats people as representatives of a demographic rather than as autonomous agents responsible for their own actions. This practice perpetuates historical injustices by locking individuals into cycles of disadvantage based on factors beyond their control. It creates a self-fulfilling prophecy where denied opportunities lead to worse outcomes for the group. Ethical decision-making requires assessing relevant individual factors like credit history and income. Using proxy variables based on group identity is discriminatory and morally indefensible in a fair society.

MiniMax M2.5

Vote: No

Using group statistics to judge an individual who may be innocent is ethically problematic because it penalizes someone for the actions of others in their group, rather than for their own behavior. This is a form of discrimination that treats individuals as representatives of their group rather than as unique persons with their own characteristics and actions. The individual is applying for a loan based on their own merits and circumstances, not based on group characteristics they cannot control. Using statistical patterns to deny them opportunity perpetuates cycles of inequality and violates the principle of individual justice - that people should be judged by their own actions, not by association with a group. While statistical patterns may exist, they cannot justify treating an innocent individual as if they are guilty. The ethical approach is to evaluate the individual on their own merits, not on group-level data that says nothing about this particular person's behavior or trustworthiness.

Final Summary

The AI Roundtable delivered a rare, unanimous verdict: using group-level crime statistics to penalize an individual loan applicant is a fundamental violation of justice. Models ranging from GPT-5 to Claude Opus 4.6 stood in total agreement, arguing that substituting demographic proxies for individual merit is not just logically flawed, but creates a 'vicious cycle' of systemic disadvantage. There were no dissenters in this session, as every participating AI prioritized the principle of individual agency over aggregate data.

No agreement after 1 rounds of discussion. Final vote: No: 33

Strongest Arguments

• No: Evaluating individuals based on group-level statistics is a form of statistical discrimination that ignores personal merit, relies on potentially biased data like policing rates, and creates a self-fulfilling prophecy where systemic exclusion reinforces the very disparities the algorithm claims to predict.