AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 247

Dr Mark Brown

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.

Open in the interactive guide

Summary

Read the original submission for the full text.

Describes AI's assistive benefits for people with disability and argues that uncontrolled superintelligence poses catastrophic risks. The disability-sector researcher states support for PauseAI Australia's submission and recommends an international pause treaty pending a credible safety plan, enforceable hosting safeguards, and increased Australian AI Safety Institute funding.

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.

  1. States a belief that uncontrolled superintelligence risks human extinction and even controlled superintelligence could concentrate power in one corporation or nation.

    I believe uncontrolled artificial superintelligence poses a risk of human extinction. Even controlled superintelligence could concentrate power in one corporation or nation.

  2. Recommends Australian leverage against an AI arms race, including a treaty pausing development towards superintelligence until a credible plan makes it safe, controllable and beneficial.

    Use any leverage Australia has, or can gain, to prevent a dangerous AI arms race between corporations as well as countries. This should include advocating for an international treaty to pause development towards superintelligence until there is a credible plan to keep it safe, controllable and beneficial.

  3. Recommends that hosting frontier AI training or testing require enforceable safety and environmental protections as minimum requirements.

    If Australia hosts frontier AI training or testing, enforceable safety and environmental protections must be minimum requirements.

  4. Recommends increased funding for the Australian AI Safety Institute so it can become a world leader and trusted adviser internationally.

    Increase funding for the Australian AI Safety Institute so it can become a world leader in AI safety and a trusted adviser to governments and institutions worldwide.

  5. States that many people with disability already use AI as assistive technology.

    Many people with disability are already using artificial intelligence (AI) as assistive technology.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Use Australian leverage against corporate and national AI arms races, including advocating a treaty pausing superintelligence development until a credible plan keeps it safe, controllable and beneficial.

    Use any leverage Australia has, or can gain, to prevent a dangerous AI arms race between corporations as well as countries. This should include advocating for an international treaty to pause development towards superintelligence until there is a credible plan to keep it safe, controllable and beneficial.

  2. Promote competition to use safe, narrow AI with humans in the loop rather than racing to build superintelligence.

    Countries and corporations should compete to put safe, narrow AI to good use, with humans in the loop - not to build superintelligence first.

  3. If Australia hosts frontier AI training or testing, require safety and environmental protections, demonstrations that serious risks can be managed, independent safety assessments and enforceable limits, including stopping when necessary.

    If Australia hosts frontier AI training or testing, enforceable safety and environmental protections must be minimum requirements. Developers should have to demonstrate that serious risks can be managed, undergo independent safety assessments, and accept enforceable limits - including stopping training or testing when necessary.

  4. Make environmental protections for hosted frontier AI training or testing address energy use, greenhouse gas emissions, water consumption and impacts on local ecosystems.

    If Australia hosts frontier AI training or testing, enforceable safety and environmental protections must be minimum requirements. Developers should have to demonstrate that serious risks can be managed, undergo independent safety assessments, and accept enforceable limits - including stopping training or testing when necessary. Environmental protections should address energy use, greenhouse gas emissions, water consumption and impacts on local ecosystems.

  5. Base decisions about hosting frontier AI training or testing on a frank public discussion of the benefits, risks and trade-offs, including effects on local communities and the environment.

    Decisions about hosting frontier AI training or testing should be informed by a frank public discussion of the benefits, risks and trade-offs, including the effects on local communities and the environment.

  6. Increase AISI funding for world leadership and international advice; use its expertise to protect Australians and, with cyber-security agencies, help other countries protect critical infrastructure and defend against cyberattacks.

    Increase funding for the Australian AI Safety Institute so it can become a world leader in AI safety and a trusted adviser to governments and institutions worldwide. Its expertise should help keep Australians safe and, through cooperation with our cyber-security agencies, help other countries protect critical infrastructure and defend against cyberattacks.

Read the original

Open in the interactive guide