This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
One thing to keep in mind is opsec.
Sometimes therapy sessions include pretty personal data.
With a regular meat-based therapist, all sorts of regulations are in place to limit how the data gathered in session can be used. Crucially, such data can sometimes not be compelled as evidence. The fact that the data is mostly in his head also makes automatic analysis more difficult.
Note that a medical professional can still call the cops on you for being a threat to yourself or others, which is likely out of the scope of current LLMs. Also note that certain faiths have a much stronger protection of data shared in confession than medical professionals both in law and professional ethics.
By contrast, assume that if you do not run your LLM locally, your conversations are stored permanently on a server without your control. From my understanding, the big AI companies do not try to facilitate anonymous payments and usage (e.g. suitable crypto-currencies and communication over TOR), as this would invite all kinds of abuse.
To keep your intimate conversations linked to your legal identity secure, at least the following would have to be true:
(a) The staff of the AI company does not read them.
(b) They don't train other AIs on them.
(c) They don't get hacked.
(d) They don't get a subpoena for e.g. 'all conversations mentioning cannibal ideation' by police.
If you aren't a minor internet celebrity like Gwern, where a ton of your text is in the corpus or a lot of people talk about you, having your data trained on is a vanishingly small concern. People forget how ridiculously compressed LLMs are compared to their training corpus, even if you spill an amount of personal info, it has little to no chance of explicitly linking it to you, let alone regurgitating it.
Certainly you shouldn't really be telling AIs things you are very concerned about keeping private, but this particular route isn't a major threat.
More options
Context Copy link
This is the problem for me. I tried a fun little game recently with my OpenAI instance, in which I've still been careful about what I write. It still had a ton of personal information from our chat contexts and was able to do a decent job figuring out pressure points in my life.
If I can't have full opsec with a virtual therapist then it's as worthless to me as the judgemental lefty who will call the cops on me if I'm sad.
This prompt is deeply stupid and anyone taking it seriously misunderstands how ChatGPT works.
Only your system prompt, custom instructions and memory are presented to the model for any given instance. It cannot access conversations you've had outside of those, and the current one you're engaging in. Go ahead, ask it. If it's not explicitly saved in memory, it knows fuck all. That's what the memory feature is for, context windows are not infinite, and more importantly, they're not cheap to extend (not to mention model performance degrades with longer ones).
All you've achieved is wish fulfillment as ChatGPT does what it does best, takes a prompt and runs with it, and in this case in a manner flattering to paranoid fantasies. You're just getting it to cold read you, and it's playing along.
Well, that's disappointing:
/images/17319872848770623.webp
More options
Context Copy link
You seem to have misunderstood what I was saying here.
It didn't guess anything that I hadn't told it, it just extrapolated from the memory of multiple chat threads, asking about useful how-to topics. There's no magic.
The point is that even with my complete avoidance of anything truly personal, a platform has valuable information. If you're spilling your guts to a virtual therapist, It's a huge vulnerability.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link