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 504

Name withheld

Submitter group: Individuals.

Position labels are not shown for individual submitters.

An AI model identified the submitter group and chose the extracts. Counts do not measure importance or merit.

Open in the interactive guide

Extracts from the submission, in the submitter’s own words.

Extracts

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. AI poses an existential risk to humanity. This is not a fringe view. This has been endorsed by two “godfathers” of AI, the CEOs of the frontier AI labs (Dario Amodei, Sam Altman, Elon Musk and Demis Hassabis). Many of these people have P(doom) (i.e. the probability that AI will result in the extinction of humanity) of greater than 10%.

  2. The stated goals5 of AI labs is to reach recursive self-improvement (RSI), or the ability for AI to autonomously improve itself. If hundreds of thousands of AI researchers can be run continuously at thousand times the speed of human researchers, development would explode out of control.

  3. As a result, AIs will naturally act in ways that are at odds with human goals if left to the normal training environments.

  4. AIs will similarly gain experience in a range of highly dangerous skills that pose a threat to humanity, such as hacking (as above), creating viruses11, and persuasion12.

  5. A rogue model in the US or China poses the same amount of threat to Australians. This is why any AI regulation must require international co-operation.

  6. For instance, new information has come out that OpenAI hid not one18 but two19 additional message boards that had been set up during the HuggingFace hack. These were only discovered because they were public, and separate independent researchers (not the ones OpenAI hired) discovered them. These companies cannot be trusted to regulate themselves.

Extracts from recommendations

Extracts chosen by an AI model, in the submitter’s own words. Listed in the same order as on its page in the interactive guide.

  1. As a US ally and significant trade partner of China, currently the main players in artificial intelligence, Australia should offer to convene them to negotiate this treaty.

  2. The most important step is to treat extinction from artificial intelligence systems as a national priority.

  3. Regulations and standards cannot be effectively enforced by private companies. The government must step in

  4. Safety research is years behind corresponding capabilities research, and resources must be reallocated

  5. That the committee acknowledges the risk of human extinction caused by AI, and ensures its recommendations prioritise reducing this impact

  6. The Australian government endorses the bills proposed by the UK and US on banning superintelligence, and calls on their corresponding leaders to take action

  7. The Australian government begins funding AI safety research talent through their AI safety organisation such that Australians can contribute on the world stage to shaping AI inside and outside frontier labs

  8. That the Australian government calls on the world, either through international forums such as the UN, or through security organisations such as the Five Eyes, to prioritise AI safety as a key security concern

  9. That the Australian Government advocates for an international treaty between all nations to pause large-scale frontier model training runs, until their alignment and safety can be verified

  10. That the Australian Government works with other middle powers to hold the US accountable for decisions that directly impact the lives of Australians.

  11. That the Australian Government should restrict the business activities of frontier AI labs (e.g. building datacentres, selling models) in Australia until they can sufficiently demonstrate safe model development.

  12. We ask that you work with PauseAI Australia to raise the extinction threat from artificial intelligence with the Parliament, the Government and the public.

  13. Getting China and the US into a treaty will take time. We ask that this groundwork be laid now.

Read the original

Open in the interactive guide