Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

KPMG pulls report on AI usage due to apparent hallucinations

techcrunch.com

Post reply on HN
Print view View original

21–30 of 37 posts

Re: KPMG pulls report on AI usage due to apparent hallucinations

#21

Post by SpicyLemonZest »

simonw wrote: post #12

This isn't the first time this has happened, either. I do not understand how these consultancies - who sell these "reports" for six or seven digit sums - continue to mess this up. It should be excruciatingly embarrassing for them. I guess nobody ever got fired for paying KPMG and friends for an expensive report that supported their priors.

The problem is that there's a lot of people running around who believe the polite fictions we tell ourselves about review processes. It's very hard to explain why it doesn't work to have someone manually clean up a sloppy AI draft without discussing the fact, which many people find unacceptable, that manual review can't catch all errors.

Re: KPMG pulls report on AI usage due to apparent hallucinations

#22

Post by tjwebbnorfolk »

simonw wrote: post #12

This isn't the first time this has happened, either. I do not understand how these consultancies - who sell these "reports" for six or seven digit sums - continue to mess this up. It should be excruciatingly embarrassing for them. I guess nobody ever got fired for paying KPMG and friends for an expensive report that supported their priors.

These six-figure reports are produced by underpaid kids in their twenties working 18 hours a day.

The purpose of paying for these reports is for executives to have someone else to blame when their idea doesn't work. It has nothing to do with the correctness of the content.

Re: KPMG pulls report on AI usage due to apparent hallucinations

#25

Post by esseph »

tjwebbnorfolk wrote: post #22
simonw wrote: post #12

This isn't the first time this has happened, either. I do not understand how these consultancies - who sell these "reports" for six or seven digit sums - continue to mess this up. It should be excruciatingly embarrassing for them. I guess nobody ever got fired for paying KPMG and friends for an expensive report that supported their priors.

These six-figure reports are produced by underpaid kids in their twenties working 18 hours a day. The purpose of paying for these reports is for executives to have someone else to blame when their idea doesn't work. It has nothing to do with the correctness of the content.

This is absolutely correct in my experience. It's solely finger pointing insurance.

Re: KPMG pulls report on AI usage due to apparent hallucinations

#26

Post by SpicyLemonZest »

e12e wrote: post #15
figassis wrote: post #13

Earlier quoted context omitted.

This is where the absolutism of let agents to 100% of the work fails. You get adversarial agents pulling all reverences into a table, they might miss some, so run this a few times. Then have another set of agents, with skills like web browsing (to verify that links actually exist, maybe that references and abstracts actually match, etc), have one engineer (or agent) write a small script to help with this (just make s…

> A human can then verify the ones with under 90% certainty. How about the author actually reads the finished report a couple of times and checks all the references? It really is the lowest bar - even lower maybe than running a spell check.

The hallucinations here (https://gptzero.me/news/investigations-kpmg/) would have passed a cursory reference check. It's easy to see when it's laid out in a table that "BNP Paribas. AI Integration: Transforming Financial Journeys. The Banking Scene, 2025." is a false citation, because the title doesn't quite match and it wrongly attributes BNP Paribas authorship to an article written about BNP Paribas by some random Belgian guy doing business as "The Banking Scene". It'd be a lot harder to see when you're skimming through browser tab 9 of 45 and see all the key words match up.

Re: KPMG pulls report on AI usage due to apparent hallucinations

#27

Post by danaris »

e12e wrote: post #15
figassis wrote: post #13

Earlier quoted context omitted.

This is where the absolutism of let agents to 100% of the work fails. You get adversarial agents pulling all reverences into a table, they might miss some, so run this a few times. Then have another set of agents, with skills like web browsing (to verify that links actually exist, maybe that references and abstracts actually match, etc), have one engineer (or agent) write a small script to help with this (just make s…

> A human can then verify the ones with under 90% certainty. How about the author actually reads the finished report a couple of times and checks all the references? It really is the lowest bar - even lower maybe than running a spell check.

How about the author actually, y'know

authors

the report?

Re: KPMG pulls report on AI usage due to apparent hallucinations

#28

Post by overfeed »

tjwebbnorfolk wrote: post #22
simonw wrote: post #12

This isn't the first time this has happened, either. I do not understand how these consultancies - who sell these "reports" for six or seven digit sums - continue to mess this up. It should be excruciatingly embarrassing for them. I guess nobody ever got fired for paying KPMG and friends for an expensive report that supported their priors.

These six-figure reports are produced by underpaid kids in their twenties working 18 hours a day. The purpose of paying for these reports is for executives to have someone else to blame when their idea doesn't work. It has nothing to do with the correctness of the content.

> These six-figure reports are produced by underpaid kids in their twenties working 18 hours a day.

That's accurate, for the first draft. Similar to big legal firms - subsequent versions are signed-off and passed up (and if revisions request, down) the hierarchy, each stratum with its own billing rate(s).

Which makes me wonder when the hallucinations got added.

Re: KPMG pulls report on AI usage due to apparent hallucinations

#29

Post by tjwebbnorfolk »

overfeed wrote: post #28
tjwebbnorfolk wrote: post #22

Earlier quoted context omitted.

These six-figure reports are produced by underpaid kids in their twenties working 18 hours a day. The purpose of paying for these reports is for executives to have someone else to blame when their idea doesn't work. It has nothing to do with the correctness of the content.

> These six-figure reports are produced by underpaid kids in their twenties working 18 hours a day. That's accurate, for the first draft. Similar to big legal firms - subsequent versions are signed-off and passed up (and if revisions request, down) the hierarchy, each stratum with its own billing rate(s). Which makes me wonder when the hallucinations got added.

Not where I used to work. Any "sign off" was some director making sure the letterhead looked right.
Post reply on HN