Lovely story:
Deloitte will provide a partial refund to the federal government over a $440,000 report that contained several errors, after admitting it used generative artificial intelligence to help produce it. The Department of Employment and Workplace Relations (DEWR) confirmed Deloitte would repay the final instalment under its contract, which will be made public after the transaction is finalised. It comes as one Labor senator accused the consultancy firm of having a “human intelligence problem”.
LLM’s hallucinating? Completely shocking:
University of Sydney academic, Dr Christopher Rudge, who first highlighted the errors, said the report contained “hallucinations” where AI models may fill in gaps, misinterpret data, or try to guess answers.
I like how we’re all skimming over the fact that they fed federal compliance data into Azure OpenAI seemingly without permission or oversight.
Accountability is non existentant at that level.
Until people who sign off on the worm are held criminally responsible, nothing will change.
I seriously hope they are off the list for future government contracts, too.
Good joke
Bitch plz… These parasites are organs of the regime
I’m going for a hot take here. The issue is not using AI (barring privacy/confidentiality), but that it wasn’t thoroughly validated.
I consider AI as a new trainee. Everything has to be validated by an experienced person.
1st year makes predictable mistakes that are easier to review.
Llm will fucking fake shit and it is way harder to spot unless you resp the entire fucking thing yourself, which a that point loses all value from cost perspective.
AI is useful as support tool for middle level and senior pros since they can catch the lies as they happen
The issue is still using AI as its sensitive data and shouldnt be given to 3rd parties.
Yup. That’s the biggest problem I see with AI. People don’t realize it’s fallible. They’re so busy wanting shortcuts and reduced work time that they’re ignoring the accuracy.
That’s part of why youre starting to see reports of AI not saving any time or money, you’re basically shifting it from the creation phase to the validation phase.
