Submission 345
Mrs Belinda Gourlay
Submitter group: Individuals.
Position labels are not shown for individual submitters.
An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.
Summary
Read the original submission for the full text.
Argues Australia's AI governance should adopt a principle of social legibility: AI presence and behaviour should be disclosed to affected people, with consent, human oversight, accountability, auditability and recourse. Also argues users should be able to choose which AI model or agent they interact with.
Key points
Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.
Says people affected by AI should understand its role well enough to decide how to engage with it.
A bystander affected by an AI system should be able to understand enough about its current role to make an informed decision about how they want to engage with it.
Proposes AI should be visible and clearly disclosed when actively involved in an interaction.
AI should be visible and able to clearly convey what mode it is operating in. When AI is actively involved in an interaction, this should be clearly disclosed.
Recommends privacy by design and clear accountability for AI decisions.
Privacy should be designed into the system, and there must be clear accountability for the decisions made by AI.
Argues people should be able to choose which AI model they interact with.
People should also be able to choose for themselves which AI model they wish to interact with.
Says people adversely affected by AI need meaningful ways to question and challenge decisions and seek recourse.
there should be meaningful ways for people to ask questions, challenge decisions and seek recourse when they are adversely affected by an AI system
Says repeated reliance on an AI system can feel relational rather than purely functional, and governance should account for that distinction.
Not everyone treats AI systems as interchangeable; many users form a reliance on a specific system through repeated use. That reliance can feel relational rather than purely functional. Communication and governance should take that distinction into account.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Let bystanders, where appropriate, object, withdraw consent or alter the interaction where an AI can listen, capture or retain information for later use.
If the AI has the capability of listening, capturing information or retaining information for later use, the bystander should, where appropriate, be able to object, withdraw consent or alter the interaction.
Consider social legibility in Australian AI governance: visible AI, clear modes of operation and disclosure when AI is actively involved in an interaction.
Australian AI Governance should consider the principle of social legibility. AI should be visible and able to clearly convey what mode it is operating in. When AI is actively involved in an interaction, this should be clearly disclosed.
Ensure a human can keep meaningful oversight and intervene or disengage the system, with consent mechanisms where personal data is collected and inclusion of people with visual, cognitive or hearing impairments.
A human should be able to maintain meaningful oversight and have the choice to intervene or disengage the system. There should be consent mechanisms, particularly where personal data is being collected, and systems should be inclusive of people with visual, cognitive or hearing impairments.
Design privacy into AI systems, with clear accountability for AI decisions and designs that make clear auditability possible.
Privacy should be designed into the system, and there must be clear accountability for the decisions made by AI. Systems should be designed so that clear auditability is possible,
Ensure meaningful avenues to question and challenge AI decisions and seek recourse when people are adversely affected by an AI system.
there should be meaningful ways for people to ask questions, challenge decisions and seek recourse when they are adversely affected by an AI system
Account in communication and governance for people’s choice of AI model and for reliance that can feel relational rather than purely functional.
People should also be able to choose for themselves which AI model they wish to interact with. Not everyone treats AI systems as interchangeable; many users form a reliance on a specific system through repeated use. That reliance can feel relational rather than purely functional. Communication and governance should take that distinction into account.
Ensure people know when AI is involved, what it does, what information it captures, who is responsible and what choices they have.
As AI moves away from clearly identifiable interfaces and becomes ubiquitous, people need to know if the AI is involved in their interaction, what exactly it is doing, what information is being captured, who is responsible for it and what choices they have.
Ensure regulation preserves responsible innovation and gives developers clear expectations for transparent systems with consent, human control and accountability.
Good regulation should not prevent responsible innovation. Developers need clear expectations. This, in turn, can increase public confidence and allow developers to design transparent systems with consent, human control and accountability.