Your comment was dead, and many of your other comments are dead (some flagged.) I vouched for this one, but you may be shadow banned, or users may be picking up on LLM tendencies in your text and flagging them.
Maybe someone in Germany needs your help with the AFD Email leak; right now journalist are going through the data; your solution sounds like something that need
This is interesting and applicable to so many domains where people want to query a model about a dataset, and have it summarise or otherwise analyse that dataset.
One of the biggest problems about LLMs as machines for analysis is that they don't respect the boundaries of a problem and instead generate base on their corpus and training as well as the actual dataset in question in the context.
It would be wonderful if we could easily strip out facts about Enron somehow from a model or just instruct it to ignore anything outside the dataset (and have the instruction respected), it is quite frustrating that these machines are so fuzzy.
I had a look at the methodology page, well done for doing the hard work of going over the results a few times to find errors/inconsistencies. There are not too many inconsistencies according to that.
Of 62,205 candidate facts, 37 quoted a sentence we couldn't find in the message
The difficulty for a business considering using this (in law for example or analysing internal business documents), is that one incorrect fact could be very damaging and it's very hard to check them all.
I completely agree. It's hard to imagine this ever being perfect both because of AI and because people make mistakes in communications all the time. In our own usage of it we've found it essential to check things that seem suspicious once they're in the company brain. But, the more important gate, and one we expect may remain longer than people would like, is that we ask people to approve changes to a company brain that are proposed. Even this isn't perfect but it is a key defense from the brain turning into slop. And it's different than this demo, which was much closer to fully automated.
Any chance you'll be open-sourcing any part of the underlying model hierarchy? A lot of us could really use a good library for deriving knowledge graphs or expert systems with cool features (such as expiration dates, like your solution has, or weighted probabilities) rather than everyone re-inventing it on their own.
You're not wrong. These load-bearing air molecules are not too energized; they are quite tranquil―and the wavelengths emitted off plants' photosynthesis organs are equally pleasing.
It's not impossible to solve. We already saw this earlier this year: LLM-generated comments were everywhere on HN, rotting this place out. Then dang implemented a no-LLM-comments policy and a filter that does a remarkably good job of automatically detecting and flagging them. It is by choice that the submissions of blatantly obviously LLM-generated content are allowed to rot out ShowHN.
Ah, Enron. The good ole days when fraudulent execs still went to jail and corporations still saw consequences of fraud.
2002, Arthur Andersen gets indicted for deliberately aiding Enron in cooking the books. 2003, Bush DoJ's attorney general binds prosecutors to effectively always settle corporate cases without convictions. So nowadays you have companies like Boeing murdering 300+ without real consequences. It's a real shame.
Certainly what it feels like... but you've got: SBF, Holmes, Madoff who have done and are doing prison time. But yea, the Boeing case stands out for sure.
A company I had private stock in just sold. The email I got saying that they wanted to get me paid ASAP was electric. Finally, for just a day or two, I was in the elite of the elite: people who are owed money and that the enforcement arm of the US government will ensure are made whole. The payday was under $100, but the feeling was priceless.
On top of defrauding the investor class as others mentioned, they had already completely lost their power at that point. So it was just taking down someone who was already down. Their companies/funds were dead already.
If Arthur Andersen hadn't been indicted, it's possible it would have continued as a company and gotten away with it, at the very least much more likely than the above three.
> The good ole days when fraudulent execs still went to jail
Well, not one guy. Lou Pai cheated on his wife with stripper Melanie Fewell who was also married. He was known in the Enron offices to enjoy Houston area strip clubs. But in a first for him, he knocked Melanie up over the course of the affair. Pai's first wife and mother of his first two children Lana Lee learned about it and started divorce proceedings. To satisfy asset division as part of their divorce terms, Pai sold $250M worth of Enron stock just months before it crashed and Enron went bankrupt. Pai then married Fewell. Pai was briefly the largest(? maybe top two?) private land owners in Colorado, and his ranchland included Culebra Peak, one of CO's 14ers.
In 2008 he settled a civil suit alleging insider trading brought by the SEC. The settlement was $31.5 and does not include an admission of the SEC's claims. He was barred from serving as an officer or director of a public company for 5 years.
Lou landed on his feet. He and Melanie are still married. They and their daughter seem to be big figures in dressage, running Canaan Ranch and the Bred in the USA Syndicate, which seems to be a syndicate of 14 breeders. Maybe someone who knows horses can explain exactly how big a deal they are; they might not be much after all.
I remember when Arthur Andersen Consulting (a software consulting firm that had split off from the accounting firm) renamed itself to "Accenture" in 2001 a few months before the Enron scandal. Although they already had split from Andersen legally, I wonder if somebody in the software firm knew what was going to happen with Enron and realized they needed to dissociate themselves with the name.
Yes, yes, I know, ctrl-f Credit Suisse and pretend it's 2005 and some crank with a tin foil ball cap is telling you about the NSA's wiretap program (with 2020 hindsight). No one ever talks about the 2019 bailouts.
> always settle corporate cases without convictions
I'm not sure that's a bad idea. You can't put a corporation in jail, but you can get their money and that also requires a lower standard of proof than a criminal conviction. In theory I could see it as a better way to recover restitution for those who suffered losses.
No, Bush's AG did not bind prosecutors to effectively always settle corporate cases without convictions. Following high-profile corporate scandals (e.g., Enron, WorldCom, Tyco), Bush established the Corporate Fraud Task Force in 2002. Over the next couple of years the DOJ obtained over 500 corporate fraud convictions or guilty pleas. They charged over 900 defendants and over 60 corporate CEOs and presidents with some type of corporate fraud crime in connection with over 400 filed cases.
It's not so much related to Enron or what a historic corporation did right or what a historic corporation did wrong, although those aspects are interesting too...
What's interesting is the use of an AI, an LLM, to create a "knowledge map" (in this case a Wiki, but it could have taken other forms) of a whole bunch of disjoint, separate, and apparently (at first glance, to the naked eye of a casual observer) disparate information from different emails at different dates and times to different parties outside of the company, the only common thread being that they were sent by employees of the same company, and for the AI/LLM to be able to extract facts and a knowledge graph ("who knew what, when") again, in the form of a Wiki (but could have been in another format), and this allows "making sense of", some "higher level viewpoint of" (at least to some better degree than all of the emails alone), again, in this case, in the form of a Wiki.
Now, this set of ideas could have some application far beyond Enron, far beyond investigations, far beyond what Enron historically did right or Enron historically did wrong...
For example, large companies and other organizations could create Wikis based on public facts in email corpuses for onboarding new employees. Not just for email interactions with outside companies (customers, suppliers, etc.) but for how different systems, different business processes used by the company work, or are supposed to work. Sort of like new employee manuals where there are entries like "if this fails, then do this...", or here's our documentation which shows that we're ISO-9000 compliant, etc. etc.
Point is, the amount of uses for something like this in a corporate setting is theoretically unlimited.
Of course, there might be issues as to what information different types of employees could read about, that is, larger corporations may wish to set access controls to different types of information, but there's no reason why an AI/LLM couldn't perform this function (with human oversight!) as well. Each of those could be ironed-out on an per-company, per-case, ad-hoc basis...
But again, the point is, the amount of potential uses for something like this in a corporate setting is theoretically unlimited...
48 comments
[ 7.6 ms ] story [ 93.9 ms ] threadOne of the biggest problems about LLMs as machines for analysis is that they don't respect the boundaries of a problem and instead generate base on their corpus and training as well as the actual dataset in question in the context.
It would be wonderful if we could easily strip out facts about Enron somehow from a model or just instruct it to ignore anything outside the dataset (and have the instruction respected), it is quite frustrating that these machines are so fuzzy.
I had a look at the methodology page, well done for doing the hard work of going over the results a few times to find errors/inconsistencies. There are not too many inconsistencies according to that.
Of 62,205 candidate facts, 37 quoted a sentence we couldn't find in the message
https://www.thisandthat.chat/enron/methodology.html
The difficulty for a business considering using this (in law for example or analysing internal business documents), is that one incorrect fact could be very damaging and it's very hard to check them all.
Unless we get some kind of verified human authentication it's nearly impossible to solve. And that in itself introduces all kinds of new problems.
Welcome to the dead internet.
At best we get this
https://xkcd.com/810/
pretty cool.
2002, Arthur Andersen gets indicted for deliberately aiding Enron in cooking the books. 2003, Bush DoJ's attorney general binds prosecutors to effectively always settle corporate cases without convictions. So nowadays you have companies like Boeing murdering 300+ without real consequences. It's a real shame.
The difference here is that they defrauded the investor class.
Too nice of a way to talk about the Epstein class.
If Arthur Andersen hadn't been indicted, it's possible it would have continued as a company and gotten away with it, at the very least much more likely than the above three.
Well, not one guy. Lou Pai cheated on his wife with stripper Melanie Fewell who was also married. He was known in the Enron offices to enjoy Houston area strip clubs. But in a first for him, he knocked Melanie up over the course of the affair. Pai's first wife and mother of his first two children Lana Lee learned about it and started divorce proceedings. To satisfy asset division as part of their divorce terms, Pai sold $250M worth of Enron stock just months before it crashed and Enron went bankrupt. Pai then married Fewell. Pai was briefly the largest(? maybe top two?) private land owners in Colorado, and his ranchland included Culebra Peak, one of CO's 14ers.
In 2008 he settled a civil suit alleging insider trading brought by the SEC. The settlement was $31.5 and does not include an admission of the SEC's claims. He was barred from serving as an officer or director of a public company for 5 years.
Lou landed on his feet. He and Melanie are still married. They and their daughter seem to be big figures in dressage, running Canaan Ranch and the Bred in the USA Syndicate, which seems to be a syndicate of 14 breeders. Maybe someone who knows horses can explain exactly how big a deal they are; they might not be much after all.
https://www.usef.org/us-equestrian-teams/athletes/natalie_pa...
https://www.reddit.com/r/Superstonk/comments/10ir695/nomura_...
Yes, yes, I know, ctrl-f Credit Suisse and pretend it's 2005 and some crank with a tin foil ball cap is telling you about the NSA's wiretap program (with 2020 hindsight). No one ever talks about the 2019 bailouts.
I'm not sure that's a bad idea. You can't put a corporation in jail, but you can get their money and that also requires a lower standard of proof than a criminal conviction. In theory I could see it as a better way to recover restitution for those who suffered losses.
https://www.justice.gov/archive/opa/pr/2004/July/04_dag_495....
It's not so much related to Enron or what a historic corporation did right or what a historic corporation did wrong, although those aspects are interesting too...
What's interesting is the use of an AI, an LLM, to create a "knowledge map" (in this case a Wiki, but it could have taken other forms) of a whole bunch of disjoint, separate, and apparently (at first glance, to the naked eye of a casual observer) disparate information from different emails at different dates and times to different parties outside of the company, the only common thread being that they were sent by employees of the same company, and for the AI/LLM to be able to extract facts and a knowledge graph ("who knew what, when") again, in the form of a Wiki (but could have been in another format), and this allows "making sense of", some "higher level viewpoint of" (at least to some better degree than all of the emails alone), again, in this case, in the form of a Wiki.
Now, this set of ideas could have some application far beyond Enron, far beyond investigations, far beyond what Enron historically did right or Enron historically did wrong...
For example, large companies and other organizations could create Wikis based on public facts in email corpuses for onboarding new employees. Not just for email interactions with outside companies (customers, suppliers, etc.) but for how different systems, different business processes used by the company work, or are supposed to work. Sort of like new employee manuals where there are entries like "if this fails, then do this...", or here's our documentation which shows that we're ISO-9000 compliant, etc. etc.
Point is, the amount of uses for something like this in a corporate setting is theoretically unlimited.
Of course, there might be issues as to what information different types of employees could read about, that is, larger corporations may wish to set access controls to different types of information, but there's no reason why an AI/LLM couldn't perform this function (with human oversight!) as well. Each of those could be ironed-out on an per-company, per-case, ad-hoc basis...
But again, the point is, the amount of potential uses for something like this in a corporate setting is theoretically unlimited...
Anyway, great article, and great idea!