The Australian Department of Home Affairs has downplayed its staff’s seven-month experimentation with ChatGPT, stating that it was primarily driven by “general curiosity” rather than for writing production code or making decisions.
The department suspended internal access to ChatGPT in May after the tool had been used experimentally since November the previous year. This decision came after concerns were raised about the usage of ChatGPT, leading to an internal questionnaire to gather information on its use.
Experimentation Details
According to the results of the questionnaire, some experiments aimed to test ChatGPT’s coding and debugging capabilities. However, Home Affairs clarified that these experiments were conducted out of curiosity and for personal awareness, without producing specific codes or scripts for production purposes.
Home Affairs asserted that ChatGPT was not used to make formal decisions on behalf of the department, nor was there evidence that departmental or classified information was included in ChatGPT searches.
Lack of Formal Register
In response to criticism from Senator David Shoebridge, who questioned the absence of a formal register for ChatGPT prompts, Home Affairs stated that such prompts did not meet the threshold to be considered a “record” unless they directly influenced a decision.
Home Affairs has clarified that its staff’s use of ChatGPT was largely driven by curiosity and experimentation rather than for practical production purposes. The department emphasized that it did not find evidence of ChatGPT being used to make formal decisions or handle classified information.
