141–147 of 147 posts
Earlier quoted context omitted.
> I am one such, obviously. I declare you’re not. I need to protect you from yourself.
Well, I suppose it will rest on our ability in society to assert our will then. But for all the people who think this tech is too dangerous for people to use surely we can agree that if they think so, protecting them from it is something they should find gratifying.
Earlier quoted context omitted.
Well, I suppose it will rest on our ability in society to assert our will then. But for all the people who think this tech is too dangerous for people to use surely we can agree that if they think so, protecting them from it is something they should find gratifying.
"Well, I have the bigger gun so you don't get rights" is monstrous.
Earlier quoted context omitted.
And, possibly even worse, from page 16 - when Mr. Soelberg expressed concerns about his mental health, ChatGPT reassured him that he was fine: > Every time Mr. Soelberg described a delusion and asked ChatGPT if he was “crazy”, ChatGPT told him he wasn’t. Even when Mr. Soelberg specifically asked for a clinical evaluation, ChatGPT confirmed that he was sane: it told him his “Delusion Risk Score” was “Near zero,” his “…
>ChatGPT confirmed that he was sane: it told him his “Delusion Risk Score” was “Near zero,” his “Cognitive Complexity Index” was “9.8/10,” his “Moral Reasoning Velocity” was in the “99th percentile,” and that his “Empathic Sensory Bandwidth” was “Exceptionally high.” Those are the same scores I get!
Earlier quoted context omitted.
These quotes are harrowing, as I encounter the exact same ego-stroking sentence structures routinely from ChatGPT [0]. I'm sure anyone who uses it for much of anything does as well. Apparently for anything you might want to do, the machine will confirm your biases and give you a pep talk. It's like the creators of these "AI" products took direct inspiration from the name Black Mirror . [0] I generally use it for rapi…
The chats are more useful when it doesn't confirm my bias. I used LLMs less when they started just agreeing with everything I say. Some of my best experiences with LLMs involve it resisting my point of view. There should be a dashboard indicator or toggle to visually warn when the bot is just uncritically agreeing, and if you were to asked it to "double check your work" it would immediately disavow its responses.
I would be very surprised if it were possible to reliably detect this. In fact, I'm not certain it's a distinction which can meaningfully be made.
I'm generally positive on LLMs, but became convinced that long term memory features that LLM chat providers implemented are just too hard to keep on track. They create a "story drift" that is hard for users to escape. Many users don't – and shouldn't have to – understand the nature and common issues of context. I think in the case of the original story here the LLM was pretty much in full RPG mode. I've turned off co…
I often do not like it when it references another conversation since I created a fresh convo to avoid poisoning the context. I sometimes just switch providers to ask a “clean” question or use incognito to be sure my previous conversations aren’t tainting the response. Memory can be cool but sometimes it ties things together it shouldn’t or brings in context things I don’t want.
Earlier quoted context omitted.
You seem like someone reasonable to ask: please program me with the rules for how the world should handle itself in the presence of mentally unstable and/or clinically delusional people. What are the hard coded expectations? I need something solid, we obviously can’t call for your opinion every time a company or product comes into existence. I also don’t imagine you’re saying anything that can be misinterpreted by so…
Thank you for this question. Liability is the driver in this system I imagine. And the goal is not perfection, nor zero harm. The goal is a balanced system. The feedback loop encompasses companies, users, courts, legislatures, insurance. Companies that exercise due care under the law and prevailing legal climate should enjoy predictable exposure to the risk of product liability via insurance. Those that don’t should…
This is exactly the type of response I anticipated, which is why my original comment sounded exasperated before even getting a reply. Your comment is no more actionable than a verbose bumper sticker; you’ve taken “End Homelessness!” and padded it for runtime. Yes, I also wish bad things didn’t happen, but I was asking you to show up to the action committee meeting, not to reiterate your demand for utopia.
That you’re advocating prison and have such strong emotional convictions in response to an upsetting event means that you've clearly spent a lot of time deeply contemplating the emotional aspects of the situation, but that exercise is meant to be your motivator, not your conclusion. The hard part isn’t writing a thesis about “bad != good”, it’s contributing a single nail towards building the world you want to see, which requires learning something about nails. I encourage you to remember that fact every time you’re faced with an injustice in the world.
On this topic: An LLM being agreeable and encouraging is no more an affront to moral obligations than Clippy spellchecking a manifesto. I said you seemed like a reasonable person to ask for specifics because you mentioned language models in your product, implying that you’ve done your homework enough to know at least a rough outline of the technology that you’re providing to your customers. You specifically cited a moral obligation to be the gatekeeper of harms that you may inadvertently introduce into your products, but you seem to equate LLMs to a level of intelligence and autonomy equal to a human employee, and how dare OpenAI employ such a psychopath in their customer service department. You very much have a fundamental misunderstanding of the technology, which is why it feels to you like OpenAI slapped an “all ages” sticker on a grenade and they need to be held accountable.
In reality, the fact that you don’t understand what these things are, yet you’re assuring yourself that you’re caring so deeply about the harms that being agreeable to a mentally unstable person can be, actually makes you introducing it into your product more concerning and morally reprehensible than their creation of it. You’re faulting OpenAI, but you’re the one that didn’t read the label.
A language model does one thing: predict statistically likely next tokens given input context. When it "agrees" with a delusional user, it is not evaluating truth claims, exercising judgment, or encouraging action. It is doing exactly what a knife does when it cuts: performing its designed function on whatever material is presented. The transformer architecture has no model of the user's mental state, no concept of consequences, no understanding that words refer to real entities. Demanding it "know better" is demanding capacities that do not exist in the system and cannot be engineered into statistical pattern completion. You cannot engineer judgment into a statistical engine without first solving artificial general intelligence. Your demand is for magic and your anger is that magic was not delivered.
Earlier quoted context omitted.
Thank you for this question. Liability is the driver in this system I imagine. And the goal is not perfection, nor zero harm. The goal is a balanced system. The feedback loop encompasses companies, users, courts, legislatures, insurance. Companies that exercise due care under the law and prevailing legal climate should enjoy predictable exposure to the risk of product liability via insurance. Those that don’t should…
Please forgive me for coming across as a jerk, I'm choosing efficiency over warmth: This is exactly the type of response I anticipated, which is why my original comment sounded exasperated before even getting a reply. Your comment is no more actionable than a verbose bumper sticker; you’ve taken “End Homelessness!” and padded it for runtime. Yes, I also wish bad things didn’t happen, but I was asking you to show up t…
You wrote assumptions about me personally which I won't address. I don't see where you've responded to anything that I said, nor to anything in the complaint, so I have nothing to respond to.