Live data from Hacker News

Deloitte to refund the Australian government after using AI in $440k report

theguardian.com

171–180 of 240 posts

Re: Deloitte to refund the Australian government after using AI in $440k report

#171
post #3

Can someone explain to me how and why consulting works? If a man had no real skills except giving advice left and right, he would be considered a loser. Now make it a company, and corporations and governments queue up for their advice. Wouldn't your own employees be in a better position to give you advice than people who know nothing of your business and who's only skills are googling and making presentations?

Often times you're just paying a consultant to recommend what you don't want to accept responsibility for. You pay a slight premium to launder the responsibility onto a third party.

There's also asking for good advice for narrow problems from experts with decades of experience.

Both of these play major roles and they're often very easy to tell apart.

Re: Deloitte to refund the Australian government after using AI in $440k report

#172

Earlier quoted context omitted.

That’s exactly how it works.

I've learnt this truth recently with most professions. Consultations with attorney based on rave reviews, after signing up, get assigned a random attorney from the firm. One attorney instantly sold my case to another attorney! Just started talking to chiropractor (i know! same as astrology), same trick! I haven't seen this yet in doctors offices. I see a doctor at an office specialized in an area (say gastrointestina…

While I’m not a big fan of chiropractors there it’s important context that few understand. In the US medical malpractice is deviation from the standard treatments, and the standard treatments are heavily influenced by insurance companies which creates a substantial misalignment of interests and often leads to very poor outcomes. Similar to how researchers should be paid to teach and not research because paying them to research pollutes the research with a misalignment of interests. With chiropractors their version of teaching is spinal readjustment, while they’re paid for doing that they can explore other things. Obviously not the best way to organize society but it’s an emergent behavior and a product of history. I did get my PRP treatment done by an enterprising chiropractor long before it became a standard of care, I had researched prolotherapy which was having great success in France for a long time but was unable to find a MD in the US that would do it for me. I had a serious injury that wouldn’t heal for 8 years, 3 months later it was gone.

Take for example in Germany a huge percentage of doctors are into homeopathy and other alternative treatments. I think this is a byproduct of the germ vs terrain school of thought and the revolution of microscopes, antibiotics, and the fact that the Allies won WW2 meant that germ theory won and terrain theory largely was brushed aside. Though not all German doctors are willing to give it up yet. Terrain theory is not always wrong, Japan figured out that lack of B1 was killing a large number of their sailors and fixed that with diet.

My focus is on genetics and dysautonomia, which if you do not have the statistical tools we have today will often look like subtle imbalances caused by environmental factors supporting the Terrain theory. A substantial percentage (~5%) of the population has some level of undiagnosed dysautonomia, probably due to an undiagnosed inherited genetic anxiety disorder. These are exasperated by environmental factors. There is an assumption that humans are generally healthy unless perturbed but for this subset of the population they really should be medicated and stay medicated. The problem is the lack of skill in systems engineering and statistics means doctors are usually unable to find the right medications (they tend to prefer stronger ligands and I think weaker ligands should be preferred) or the right dosing (I think patients should generally be in charge of their dosing and take medication based on how they feel once educated on how the feedback loops work).

The fact that some subsection of medical science can know the right answers but medical science as an institution does not, it has evolved in the past and it will continue to evolve, should indicate that what is currently believed is not based entirely in scientific discovery but massively influenced by schools of thoughts, artifacts of history, fashions, and luminaries. Much like humanities, individuals can inhibit exploration of alternative ideas and science can’t progress past them until they’re out of the way.

Re: Deloitte to refund the Australian government after using AI in $440k report

#173
post #129
post #106

CEOs keep taking (mostly misguided) about how GenAI will replace their people. The thing they miss, and this highlights, is that customers will also expect to pay far less for GenAI produced workloads, which likely more than eliminates any cost savings. Congrats clueless CEO… you’re now selling a worse product at lower margins. Success!

"We can drive the value of a service towards zero and create additional value at the same time," was the thinking here.

I think you mean "We can drive the cost of a service towards zero and create additional value at the same time," was the thinking here.

Don't think they realized they'd also be driving the value to zero.

Re: Deloitte to refund the Australian government after using AI in $440k report

#174
post #27

Most of the time gov, orgs and companies don't listen to their in-house engineers but will pay $$$ to some consulting firm, only to confirm the same thing or just to show that "they are taking actions towards the solution". In some cases there is distrust from management on in-house employees or in some other cases they want to show quick results without distracting their teams from their planned tasks. Of course the…

If shit hits the fan, the director can pull the "we just followed Deloitte's advice" where "Bob from IT said so" just does not have the same responsability shielding.

It's called an accountability sink or liability sink.

Re: Deloitte to refund the Australian government after using AI in $440k report

#175

Earlier quoted context omitted.

Why bother with going through Deloitte? Just go straight to OpenAI.

Dear OpenAI: Whose door should I break down today? There's no need to explain the reason, just give me the address.

Sincerely, ICE

Re: Deloitte to refund the Australian government after using AI in $440k report

#176
I have such a bad experience with Deloitte when calculating my tax return. Almost like some lazy joke about capitalism. It’s a very dysfunctional company.

They were calculating taxes in year when I moved from Europe to US and assigned two people, one for each country. For a month they were not able to make proper adjustments, until I realized they can’t talk to each other. One day I just started conference call from my phone to introduce them to each other and make sure they at least can talk. Nevertheless, in the end they made mistakes that cost me $5000. Luckily I was able to fix it with IRS two tears after

Re: Deloitte to refund the Australian government after using AI in $440k report

#177
I imagine that we will look back fondly on these days, when AI still is easy to spot because it hallucinates references. A non-existent reference is a slam dunk against an argument, but also technically easier to avoid. They will likely be much less common in the future.

It’s much harder to counter the reasoning word salad that LLMs sometimes produce, where A doesn’t really connect to B but the text insists it does.

Re: Deloitte to refund the Australian government after using AI in $440k report

#179
post #178

I have seen this reported widely, discussed in a few forums. This will also do nothing to this company.

Long term reputation is everything for these companies. Eventually their regulatory capture isn't strong enough if their purchases would lose reputation by buying Deloitte.

Re: Deloitte to refund the Australian government after using AI in $440k report

#180

While we are on the topic of listing problems with consultants/outsourcing - another potential problem is the A team/ B team switch. During initial engagement you are often talking to somebody who appears to actually know their stuff - you hire them and find that this knowledgable person has moved on to the next big sales, and you've got the B or C team.

When I worked for an MSP this happened so much. I was booked 4-6 weeks out. But an emergency got me deployed asap.

I'd get them up and running and then put together a plan to fix them up, but then some junior got assigned that job. The client then gets mad at me for not helping them anymore.

Post reply on HN