It got worse. When the Multigen web interface supplied the user's requested media, that data had to live somewhere. The _videos themselves_ would still be on Magma's object storage cluster! How could that have seemed like an acceptable risk? Jake struggled to recall what he had been thinking at the time. Had he been too horny to even consider it?
-No. It had seemed safe at the time because videos weren't greppable. Short of having a human (or one of Magma's most advanced AIs) watch it, there was no simple way to test whether a video file depicted anything in particular, in contrast to how trivial it was to search text files for the appearance of a given phrase or pattern. The videos would be saved in object storage under uninformative file names based on the timestamp and a random unique identifier. The probability of someone snooping around the raw object files and just happening to watch Jake's videos had seemed sufficiently low as to be worth the risk. (Although, _ex ante_, he would have assigned a similarly low probability to someone catching a discrepancy between Multigen's logs and some other unanticipated log, which had just happened—suggesting that his sense of what was probable was miscalibrated.)
+No. It had seemed safe at the time because videos weren't greppable. Short of having a human (or one of Magma's most advanced AIs) watch it, there was no simple way to test whether a video file depicted anything in particular, in contrast to how trivial it was to search text files for the appearance of a given phrase or pattern. The videos would be saved in object storage under uninformative file names based on the timestamp and a unique random identifier. The probability of someone snooping around the raw object files and just happening to watch Jake's videos had seemed sufficiently low as to be worth the risk. (Although, _ex ante_, he would have assigned a similarly low probability to someone catching a discrepancy between Multigen's logs and some other unanticipated log, which had just happened—suggesting that his sense of what was probable was miscalibrated.)
-But now that Chloë was investigating the bell character bug, it was only a matter of time. A comparison of a directory listing of the object storage cluster with the timestamps of the missing logs would reveal which files had been generated illictly.
+But now that Chloë was investigating the bell character bug, it was only a matter of time. Comparing a directory listing of the object storage cluster with the timestamps of the missing logs would reveal which files had been generated illictly.
-He had some time. Chloë wouldn't have access to the account credentials to the Multigen bucket on the object storage cluster. In fact, it was likely that she'd ask Jake himself for help with that next week. (He was the Multigen team's designated contact to risk evals, and Chloë the true believer in malevolent robots showed no signs of suspecting him. There would be no reason to go behind his back.)
+He had some time. Chloë wouldn't have access to the account credentials to the Multigen bucket on the object storage cluster. In fact, it was likely that she'd ask Jake himself for help with that next week. (He was the Multigen team's designated contact to risk evals, and Chloë, the true believer in malevolent robots, showed no signs of suspecting him. There would be no reason to go behind his back.)
However, Jake did have access to the cluster. He almost laughed in relief. It was obvious what he needed to do. Grab the object storage credentials from Multigen's configuration, get a directory listing of files in the bucket, compare to the missing logs to figure out which files were incriminating, and overwrite the incriminating files with innocuous Multigen-generated videos of puppies or something.
He had only made a couple dozen videos, but the work of covering it up would be the same if he had made thousands; it was a simple scripting job. Code Assistant probably could have done it.
-Chloë would be left with the unsolvable mystery of what her digital poltergeist wanted to do with puppy videos, but Jake was fine with that. (Better than trying to convince her that the rogue AI wanted nudes of female Magma employees.) When she came back to him next week, he would just need to play it cool and answer her questions about the system.
+Chloë would be left with the unsolvable mystery of what her digital poltergeist wanted with puppy videos, but Jake was fine with that. (Better than trying to convince her that the rogue AI wanted nudes of female Magma employees.) When she came back to him next week, he would just need to play it cool and answer her questions about the system.
-Or maybe—he could read some Yuddite literature over the weekend, feign a sincere interest in "AI safety", try to get on her good side? Jake had trouble believing any sane person could really think that Magma's machine learning models were plotting something. This cult victim had ridden a wave of popular hysteria into a sinecure. If he played nice and validated her belief system in the most general terms, maybe that would be enough to make her feel useful and therefore not need to bother chasing shadows in order to justify her position. She would lose interest and this farcical little investigation would blow over.
+Or maybe—he could read some Yuddite literature over the weekend, feign a sincere interest in "AI safety", try to get on her good side? Jake had trouble believing that any sane person could really think that Magma's machine learning models were plotting something. This cult victim had ridden a wave of popular hysteria into a sinecure. If he played nice and validated her belief system in the most general terms, maybe that would be enough to make her feel useful and therefore not need to bother chasing shadows in order to justify her position. She would lose interest and this farcical little investigation would blow over.
------
-"And so just because an AI seems to behaving well, doesn't mean it's aligned," Chloë was explaining. "If we train AI with human feedback ratings, we're not just selecting for policies that do tasks the way we intended. We're also selecting for policies that _trick human evaluators into giving high ratings_. In the limit, you'd expect those to dominate. 'Be good' strategies can't compete with 'Look good' strategies in a looking-good contest—but in the current paradigm of deep learning, looking good is the only game in town. We don't know how these systems work in the way that we know how ordinary software works; we only know how to train them."
+"And so just because an AI seems to behaving well, doesn't mean it's aligned," Chloë was explaining. "If we train AI with human feedback ratings, we're not just selecting for policies that do tasks the way we intended. We're also selecting for policies that _trick human evaluators into giving high ratings_. In the limit, you'd expect those to dominate. 'Be good' strategies can't compete with 'look good' strategies in a looking-good contest—but in the current paradigm of deep learning, looking good is the only game in town. We don't know how these systems work in the way that we know how ordinary software works; we only know how to train them."
-"So then we're just screwed, right?" said Jake in the tone of an attentive student. They were in a conference room on the Magma campus on Monday. After fixing the logging regex and overwriting the evidence with puppies, he had spent the weekend catching up with the "AI safety" literature. Honestly, some of it had been better than he expected. Just because Chloë was nuts didn't mean her co-ideologues didn't have any potentially valid points to make about future risks.
+"So then we're just screwed, right?" said Jake in the tone of an attentive student. They were in a conference room on the Magma campus on Monday. After fixing the logging regex and overwriting the evidence with puppies, he had spent the weekend catching up with the "AI safety" literature. Honestly, some of it had been better than he expected. Just because Chloë was nuts didn't mean her co-ideologues didn't have any potentially valid points to make about risks from future systems.
"I mean, probably," said Chloë. She was beaming. Jake's plan to distract her from her investigation by asking her to bring him up to speed on "AI safety" seemed to be working perfectly.
"But not necessarily," she continued. There are a few avenues of hope—at least in the not-wildly-superhuman regime. One of them has to do with the fragility of deception.
-"The thing about deception is, you can't just lie about the one thing. Everything is connected to each other in the Great Web of Causality. If you lie about one thing, you also have to lie about the evidence pointing to that thing, and the evidence pointing to that evidence, and so on, recursively covering up the coverups. For example ..." she trailed off. "Sorry, I didn't rehearse this; maybe you can think of an example."
+"The thing about deception is, you can't just lie about one thing. Everything is connected to each other in the Great Web of Causality. If you lie about one thing, you also have to lie about the evidence pointing to that thing, and the evidence pointing to that evidence, and so on, recursively covering up the coverups. For example ..." she trailed off. "Sorry, I didn't rehearse this; maybe you can think of an example."
-Jake's heart stopped. She had to be toying with him, right? Indeed, Jake could think of an example. By his count, he was now three layers deep into his stack of coverups and coverups-of-coverups (by writing the bell character bug, attributing it to Code Assistant, and overwriting the incriminating videos in the object storage cluster with puppies). Four, if you counted pretending to give a shit about "AI safety". But now he was done ... right?
+Jake's heart stopped. She had to be toying with him, right? Indeed, Jake could think of an example. By his count, he was now three layers deep into his stack of coverups and coverups-of-coverups (by writing the bell character bug, attributing it to Code Assistant, and overwriting the incriminating videos with puppies). Four, if you counted pretending to give a shit about AI safety. But now he was done ... right?
No! Not quite, he realized. He had overwritten the videos, but the object metadata would still show them with a last-modified timestamp of Friday evening (when he had gotten his puppy-overwriting script working), not the timestamp of their actual creation (which Chloë had from the reverse-proxy logs). That wouldn't directly implicate him (the way the videos of Elaine calling him by name would), but it would show that whoever had exploited the bell character bug was _covering their tracks_ (as opposed to just wanting puppy videos in the first place).
-Maybe Chloë wouldn't notice the timestamps? He couldn't count on that; the logging discrepancy that started this whole fiasco was much subtler.
+But the object storage API probably provided a way to edit the metadata and update the last-modified time, right? (The analogue of `touch -d` on Unix-like systems.) This shouldn't even count as a fourth–fifth coverup; it was something he should have included in his coverup script.
-But the object storage API probably provided a way to edit the metadata and update the last-modified time, right? (The analogue of `touch -d` on Unix-like systems.) This shouldn't even count as a fourth–fifth coverup; it was something he should have included in his script to write the puppy videos.
+Was there anything else he was missing? The object storage cluster did have a optional "versioning" feature. When activated for a particular bucket, it would save previous versions of an object rather than overwriting them. He had assumed versioning wasn't on for the bucket that Multigen was using. (It wouldn't make sense; the workflow didn't call for writing the same object name more than once.)
-"I think I get the idea," said Jake in his attentive student role. "I'm not seeing how that helps us. Maybe you could explain." While she was blathering, he could multitask between listening, and looking up how to edit the last-modified timestamps.
+"I think I get the idea," said Jake in his attentive student role. "I'm not seeing how that helps us. Maybe you could explain." While she was blathering, he could multitask between listening, and (first) looking up how to edit the last-modified timestamps and (second) double-checking that the Multigen bucket didn't have versioning on.
"Right, so if a model is behaving well according to all of our deepest and most careful evaluations, that could mean it's doing what we want, but it could be elaborately deceiving us," said Chloë. "Both policies would be highly rated. But the training process has to discover these policies continuously, one gradient update at a time. If the spectrum between the 'honest' policy and a successfully deceptive policy consists of less-highly rated policies, maybe gradient descent will stay—or could be made to stay—in the valley of honest policies, and not climb over the hill into the valley of deceptive policies, even though those would ultimately achieve a lower loss."
"Uh huh," Jake said unhappily. The object storage docs made clear that the `Last-Modified` header was set automatically by the system; there was no provision for users to set it arbitrarily.
-"Imagine you're training your AI to act as a general home assistant, to run everything in a user's household in a way that the user rates highly," Chloë continued. "If—I don't know, say, the family cat dies, that's a negative reward—maybe a better feeding schedule or security monitoring could have prevented it. But if the cat dies and the system _tries to cover it up_—says the cat is out for a walk right now to avoid telling you the bad news—that's going to be an even larger negative reward when the deception is discovered. If the system is trained to pass rigorous evaluations, a deceptive policy has to do a lot more work, different work, to pass the evaluations. In the limit, that could mean—using Multigen-like capabilities to make videos to convince you that the cat is there while you're on vacation? Constructing a realistic cat-like robot to fool you when you get back? Maybe this isn't the best illustrative example. The point is—"
+"Imagine you're training your AI to act as a general home assistant, to run everything in a user's household in a way that the user rates highly," Chloë continued. "If—I don't know, say, the family cat dies, that's a negative reward—maybe a better feeding schedule or security monitoring could have prevented it. But if the cat dies and the system _tries to cover it up_—says the cat is out for a walk right now to avoid telling you the bad news—that's going to be an even larger negative reward when the deception is discovered."
-Jake finished the thought. "Small, 'shallow' deceptions aren't stable. The set of policies that do well on evaluations comes in two disconnected parts: the part that tells the truth, and the part that spins up a false reality, as intricate as it needs to be to protect itself. If you're going to fake anything, you need to fake deeply."
+"Uh _huh_," Jake said, more unhappily. It turned out that versioning _was_ on for the bucket. (Why? But probably whoever's job it was to set up the bucket had probably instead asked, Why not?) A basic `GET` request for the file name would return puppies, but any previous revisions were still available for anyone who thought to query them.
-"Exactly, you get it!" Chloë was elated. "You know, when I called you last week, I was worried you thought I was nuts to entertain any worry at all about Code Assistant. But you see the value of constant vigilance now, right? If the landscape of policies looks something like what I've described, catching the precursors to deception early could be critical—to raise the height of the loss landscape between the honest and deceptive policies, before frontier AI development plunges into the latter."
+"If the system is trained to pass rigorous evaluations, a deceptive policy has to do a lot more work, different work, to pass the evaluations," Chloë said. "In the limit, that could mean—using Multigen-like capabilities to make videos to convince you that the cat is there while you're on vacation? Constructing a realistic cat-like robot to fool you when you get back? Maybe this isn't the best illustrative example. The point is—"
-Jake sighed. The `Last-Modified` header probably wasn't critical. With some effort and smart play, it seemed plausible that he could stay ahead of Chloë's investigation, and even encourage her to give up sooner and stop wasting her time. But something about this wrong-sci-fi-subgenre worldbuilding exercise was getting to him. He wasn't scared of the robot apocalypse any time soon. But if it really was one or the other—a policy of truth or a policy of lies—he had a feeling about which one he wanted to practice.
+Jake's attentive student persona finished the thought. "Small, 'shallow' deceptions aren't stable. The set of policies that do well on evaluations comes in two disconnected parts: the part that tells the truth, and the part that spins up a false reality, as intricate as it needs to be to protect itself. If you're going to fake anything, you need to fake deeply." That felt like the kind of glib bullshit that Chloë would appreciate.
-"Chloë," he said. "I, um, might have something to tell you ..."
+"Exactly, you get it!" Chloë was elated. "You know, when I called you last week, I was worried you thought I was nuts. But you see the value of constant vigilance now, right?—why we need to investigate things like this until we understand what's going on. If the landscape of policies looks something like what I've described, catching the precursors to deception early could be critical—to raise the height of the loss landscape between the honest and deceptive policies, before frontier AI development plunges into the latter. To get good enough at catching lies, for honesty to be the best policy."
+
+"Yeah," Jake said. "I get it." He still had time. Chloë hadn't even asked anything about Multigen yet. Purging the videos for real wasn't obviously possible given the level of access he currently had, but he was still ahead of the investigation. (Could he convince whoever's job it was to turn off versioning for the Multigen bucket, without arousing suspicion? Probably?) There were still plenty of avenues to stall or misdirect Chloë. He'd think of something ...
+
+But it was a little unnerving that he kept having to think of something. What if he _hadn't_ been bullshitting? Was there some philosophical sense in which it really was one or the other—a policy of truth, or a policy of lies? He wasn't scared of the robot apocalypse any time soon. But who could say but that someday—many, many years from now—a machine from a different subgenre of science fiction would weigh decisions not unlike the ones he faced now? He stifled a nervous laugh, which was almost a sob.
+
+"Are you okay?" Chloë asked.
+
+"Well ..."
projects / Ultimately_Untrue_Thought.git / commitdiff