When a consultancy uses an AI consultant

Whom do you blame when a $290,000 gov report is riddled with errors? The consultancy firm that outsourced its prep to a chatbot? The chatbot that hallucinated through policy recommendations to introduce multiple errors? Or the gov that paid top dollar for this stitch job? Consultancy biggie Deloitte had decided Azure OpenAI GPT-4o to be its new intern - except this intern doesn't sleep, doesn't take leave, doesn't need 'one more day' to finish the job. But the clue to what holds for the future of human consultant-AI consultant co-consultancy may lie in the fact Deloitte has decided to return part of its consultation fee to the Aussie gov. Which means, it does keep some money for work it made its slave AI to do.

This AI outsourcing throwing up mistakes and left uncorrected by Deloitte can go one of two ways. Deloitte may be signalling: 'So this is what happens if you lean on AI too much!' (Read: Peeps, you will need human smarties like us till the end of time.) Or, it could be a shot in its own foot, with clients now deciding it's time to take out the middle man - no, not AI, but consultancy firms, and then have some of their own folks fact-check. As for reports coming out with glitches, we have a new dog in town to blame for eating the homework: AI. The real winner? The invoice printer. It'll never make a 'mistake'.