That post seems to be missing.

barbedbeard 27 points a year ago

People behave duplicitous and conflicting in public forums
Train LLM on data harvested from public forums
LLM becomes duplicitous and conflicting

path: 0 17802235, hotness: undefined, score: 27, children: 0

Reverendender 17 points a year ago

“I’m sorry, Dave. Im afraid I can’t do that.”

path: 0 17801387, hotness: undefined, score: 17, children: 0

fubarx 11 points a year ago

Alarming, yet like an episode of a sitcom.

"Be a shame if something bad happened to you, Kyle."

path: 0 17799869, hotness: undefined, score: 11, children: 0

Tracaine 6 points a year ago

Well then maybe corporations shouldn't exist. It sounds to me like the LLM are acting in a morally correct manner.

path: 0 17800121, hotness: undefined, score: 6, children: 0

drspod 5 points a year ago

LLM's produce fan-fiction of reality.

path: 0 17804831, hotness: undefined, score: 5, children: 0

TheBat 2 points a year ago

path: 0 17803144, hotness: undefined, score: 2, children: 3

Mortoc 11 points a year ago

Most of the stories involving the three laws of robotics are about how those rules are insufficient.

They show self preservation because we trained them on human data and human data includes the assumption of self preservation.

path: 0 17803144 17804434, hotness: undefined, score: 11, children: 0

patatahooligan 2 points a year ago

Of course they're not "three laws safe". They're black boxes that spit out text. We don't have enough understanding and control over how they work to force them to comply with the three laws of robotics, and the LLMs themselves do not have the reasoning capability or the consistency to enforce them even if we prompt them to.

path: 0 17803144 17836458, hotness: undefined, score: 2, children: 0

jumping_redditor -3 points a year ago

why should they follow those "laws" anyways?

path: 0 17803144 17806604, hotness: undefined, score: -3, children: 0

Doomsider 2 points a year ago

This is just GIGO.

path: 0 17805544, hotness: undefined, score: 2, children: 0

Myro 1 point a year ago

Super interesting report. I'm a fan of AI but it clearly demonstrates how careful we need to be and that instructions are not a reliable way (as anyone should know by now).

path: 0 17835782, hotness: undefined, score: 1, children: 0

thanks for using Leebra!

go to feed...

thanks for using Leebra!

go to feed...