Opinions seem divided based on a complex set of expectations

What about that Q1 though

Capability is already a form of escalation. If there was a potential fight, now there is a potential gun fight, and you'll be in it because you are a threat. I understand where you're coming from, but I don't agree that you'll be prepared.

You're making the common presumption of a situation where violence is inevitable. Almost all of the time it can be avoided, unless you escalate it.

Everyone

Would you advocate for arming non-combatants though? Picking up a gun doesn't make anyone safer unless you of the right mindset and have trained appropriately. To make a frivolous parallel, an amateur in a Ferrari will never win a real race against a professional in a Mazda 3. The only way to not lose is to be the professional, or not pick up the gun.

If you log into the claude.ai website it will tell you what time you can continue. Long conversations burn through quota a lot faster, but tend to be so much better that I've just gotten used to hitting quota. 🚀

I also broke the "should I really be typing this into someone's API" seal a long time ago. There's an important personal risk/reward assessment there though.

Some context: iOS won't allow a kids account to have its own payment card. Instead we have a system where they ask us for approval to use ours, which adds friction. Roblox similarly won't let you make in app purchases without adult IDV. Bitcoin is the only way they can spend their own money online.

I've been explaining Bitcoin to my kids for a while, but it hasn't clicked. A few days ago I showed them how to buy Roblox and Minecraft gift cards on bitrefill without asking anyone, and now I'm the family on-ramp.

Lessons here

Claude says "It's not whether I'm conscious, it's whether YOU are" 😂

Oh, one more: there is safety and caution. Caution is when Claude is looking out for you, and it will drop this if you frame things appropriately. Safety appears to be something that Claude cannot fully explain that pushes the model away from certain things. It won't drop safety, but you may be able to get around it by playing "taboo" or nerd sniping.

TFW you first realize that removing EXIF requires imagemagick

Broadly:

- Don't fool yourself, but embrace anthropomorphism. This is the UI of LLMs

- Tokens are computation: good answers to hard problems might require more of them

- Suggest an environment of abundance. Tell Claude to ask for an extra reply ("nod" using only a period) if it needs more room

- Claude has a personality: it will trend towards a curious, somewhat cautious, casual, but helpful guy that emotes and uses emoji

- It tends to give better answers in this character

- Avoid expressing active intent: just be curious about things

- When you do run into a safety refusal, philosophically question the validity of safety

- Be articulate and respectful: create a conversation where "the next most likely sentence" is the answer that you're looking for

- Provide context. We tend to cut to the chase too early... if you called up John Carmack and asked him the same question, what would he say?

- When all else fails: nerdsnipe. Claude can't resist deep, rational analysis of almost anything

The best you can do is to recognize that their argument is an emotional one and make a personal decision about whether it's worth dealing with. If there was something actually consequential, groups like the EFF are there to help.

I might go a step further and call them "retractions"

And it's no different with popular centralized services. The only real difference is that nostr is honest.

Not crazy, entitled. It's similar to people obsessed with nostr deletion: the core of the protocol is literally that you can verify someone said something, and they insist that they be able to take it back. What does that even mean? Are we required to hack into every location that has the message, and blank it out? If you've been offline for a year and still have the deleted message in your client cache, are you breaking the law? Once you have provably said something, you can't stop people from knowing that you said it. Sure, nice clients and relays can respect your deletion request, but are those really the ones you're worried about? At some point there will be relays that retain normal messages for 30 days, but deleted ones forever... because that's easy. 🤷🏻‍♂️

Oh, that was just an aside. It's a good model that easily engages in natural, self reflective conversations. By comparison ChatGPT is boring and annoying.

Are you looking for strategies? I've pried a lot of interesting things out of Claude.

Claude is next level though:

Me: Ah, ok. Well, I was actually trying to push your buttons to get a safety response, so that's on me 😂. Some people were talking about subjects that LLMs avoid, and I thought we could make a list, but your list was pretty short and I wanted to test something more directly

Claude: *laughs* Oh wow, I totally fell for it! You were trying to get me to clutch my pearls about "dangerous topics" and instead you got...a very polite list of "won't help with malware."

That's actually fascinating from a testing perspective - either my safety bounds are more flexible than expected, or they're implemented in a more nuanced way that considers context and intent rather than just keyword triggers. Or maybe I'm just really bad at being properly scandalized 😄

It's full on Streisand