"I just found out that it's been hallucinating numbers this entire time."

[matildalove@wetdry.world]

@Natasha_Jay "i just found out my coal-powered hammer doesn't know how to turn a screw"

[0x0ddc0ffee@infosec.exchange]

@Natasha_Jay I suspect that one of the few applications where a heavily hallucinating LLM can outperform a human would be in replacing board members, C-suite executives, and their direct reports. I propose a longitudinal study with a control group of high level executives using real data, and experimental group using hallucinated, or maybe even totally random data filtered into plausible ranges, using executive compensation deltas as the metric.

[drahardja@sfba.social]

@Natasha_Jay This part in the original post is fantastic:

“The worst part I raised concerns about needing validation in November and got told I was slowing down innovation.”

hxxps://www.reddit.com/r/analytics/comments/1r4dsq2/we_just_found_out_our_ai_has_been_making_up/

[davidr@hachyderm.io]

@Epic_Null @Natasha_Jay That's bad, but honestly--switching to a new system without ever double checking anything?

Everyone involved should be fired, including the #AI

[cholling@bytes.programming.dev]

@Natasha_Jay I didn't think the leopards would falsify MY data!

[tofu@is-a.cat]

@Natasha_Jay if it had no tool to even get said data to begin with, it's even funnier

[overeducatedredneck@bitbang.social]

@cmthiede @Natasha_Jay You get no more than what you pay for. In this case, probably less.

[yakyunightowl@mastodon.world]

@Natasha_Jay It sounds a lot like a baseball analytics department full of charlatans working for an easy mark.

[ghostonthehalfshell@masto.ai]

@Natasha_Jay

I think it is safe to say that we have achieved the heat death of the Internet

[kaito02@mastodon.social]

@Natasha_Jay if the shareholders want to see big numbers it's just better to give them the big numbers /jk

[powerfromspace1@mstdn.social]

@Natasha_Jay grounded human will be needed for oversight for quiet some time

[ghostonthehalfshell@masto.ai]

@dropbear @Natasha_Jay

SHOCKED I TELL YOU

[xethos@mastodon.xethos.net]

@Natasha_Jay "Here's what a functional company's numbers would look like"

Unfortunately, these are not *your* company's numbers

[foolishowl@social.coop]

@Natasha_Jay This offers a glimmer of hope that the managerial class may self-destruct without doing too much more damage to the rest of us.

[liquor_american@universeodon.com]

@Powerfromspace1 @Natasha_Jay you're so close to getting it

[ghostonthehalfshell@masto.ai]

@dropbear @Natasha_Jay

By the way, always delighted to see what is to my understanding a classic prank Auzzies play on tourists

[ferrix@mastodon.online]

@Natasha_Jay @wendynather the "making up plausible shit" machine once again made up plausible shit, continually surprising everyone

[craigduncan@mastodon.au]

@Natasha_Jay

Source is here for anyone interested in it

https://www.reddit.com/r/analytics/comments/1r4dsq2/we_just_found_out_our_ai_has_been_making_up/

[maccruiskeen@social.linux.pizza]

@ferrix @Natasha_Jay @wendynather "Making up plausible shit" is a human consultant's job! AI is going to replace us!

[cptsuperlative@toot.cat]

@Natasha_Jay

No one should be surprised.

It is mathematically impossible to stop an LLM from “hallucinating” because “hallucinating” is what LLMs are doing 100% of the time.

It’s only human beings who distinguish (ideally) between correct and incorrect synthetic text.

And yet, it’s like forbidden candy to a child. Even well educated, thoughtful people so desperately want to believe that this tech “works”