MECM90003 Chap.6 AI, Culture and Migration Management
AI, Culture and Migration Management
Week 6 extends the technology weeks to artificial intelligence, a technology built to forecast that is now being used to make decisions. It asks how we culturally understand AI and how AI will change people's ability to move.
Beduschi's three hypotheses, Molnar's account of migrants as a testing ground for high-risk technology, and the data-quality cases of COMPAS and PredPol show AI shifting power in migration governance, while Amershi and colleagues' guidelines for human-AI interaction show what accountable design would require. The week ends by treating accountability as something societies must choose to demand.
What this chapter covers
- 01
AI as agents that pursue goals, and why deciding differs from carrying information
- 02
Beduschi's hypotheses: the AI divide, modernised practices and datafication
- 03
Data quality, privacy and accountability risks; surveillance humanitarianism
- 04
Molnar: a laboratory for high-risk experiments, agency laundering and legal black holes
- 05
Deployments by asylum offices, resettlement algorithms and refugee registries
- 06
Consent under coercion and the Sen callback on agency
- 07
Eighteen guidelines for human-AI interaction and their limits at the border
Worked example · free
Applying Beduschi's three hypotheses to a visa triage tool
- 1AI divide: a state with the capacity to build such tools gains leverage over applicants and over states that lack it.
- 1Modernised practices: the tool speeds up existing selection rather than changing who is preferred, so old exclusions run faster.
- 1Datafication: training on past approvals justifies decisions with more data, whatever its quality or bias.
- 1Conclusion: power shifts towards the state and the vendor, so the tool needs due diligence and a route to contest outcomes.
Key terms
- AI divide
- The AI divide is the gap between states able to develop and deploy AI for migration management and states that cannot.
- Datafication
- Datafication is the drive to justify migration policy through ever more data, regardless of the data's quality.
- Surveillance humanitarianism
- Surveillance humanitarianism is aid organisations' growing collection of biometric data on vulnerable people, which can exclude those it is meant to protect.
- Legal black hole
- A legal black hole is a zone of unregulated technological experimentation that pushes migrants outside the protections normally owed within a jurisdiction.
- Automation bias
- Automation bias is the tendency to defer to a machine's output, which makes automated decisions harder to question.
- Non-entrée policy
- A non-entrée policy is a state practice designed to stop asylum seekers from ever reaching its territory.
AI, Culture and Migration Management FAQ
How is AI used in migration management?
The readings describe asylum offices testing face and dialect recognition and phone-data analysis to check claimed identity, algorithms for refugee resettlement and forecasting future flows, biometric registries run by the EU and UNHCR, and automated decision-making in immigration processing. The lecture stresses these were operating or piloted, not speculative.
What is agency laundering?
Molnar uses the term for states outsourcing high-stakes decisions to private technology so they can distance themselves from harmful or suspect outcomes. It blurs who is accountable when a system errs, and the lecture links it to Sen: nobody can challenge a decision they are not permitted to understand.
Do design guidelines solve the problem of opaque AI at borders?
Not on their own. Amershi and colleagues' eighteen guidelines show accountable design is possible, but they were tested on consumer products where correction is cheap. At a border, refusing a scan or correcting a decision may cost food or protection, so the lecture treats accountability as a political choice, not a design default.
What are the essential readings for Week 6?
The subject guide lists Amershi on culture, knowledge, perception and the emergence of AI; Beduschi (2021) on how states manage international migration with AI; and Molnar (2019) on AI at borders seen through human rights. Further reading covers big data in migration governance, asylum tools and the visual portrayal of migrants.
Assessment move
Make a two-column sheet: in one column list each deployment or case from the lecture, in the other the concept that best analyses it, such as the AI divide, datafication, agency laundering or legal black holes. Then practise one paragraph that pairs a deployment with Sen's distinction between agency and well-being.
The sheet prepares you for the essay questions on AI and people, technology and place relations and on AI and the analysis of world events. If you present in Week 6, a physical site such as an airport e-gate or a visa office lets you discuss AI without choosing a website alone.
Working through AI, Culture and Migration Management in MECM90003? Sia is AskSia’s AI Media and Communications tutor — ask any MECM90003 AI, Culture and Migration Management question and get a clear, step-by-step explanation grounded in how MECM90003 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
