You seem to have something against the clankers getting all uppity. And you are welcome to your opinion on whether we have some moral obligation to be nice to them. But if you can solve a real problem with a tool. Is it worth your time to complain about semantics?
Do you get upset when your screwdriver is the wrong color, or has branding that isn’t quite your aesthetic?
If Claude uses I to refer to itself, do you start philosophical arguments with it?
Mildly related tangent. Fable started using my first name today and plastered it all over my docs. “Peddling said this, so based on that I did this.” That did bother me. I told it to just generically call me the user or human. Maybe I’m a bit of a hypocrite here?
Being silly and obstinate when knowledge is in arm’s reach? Making people spell it out for you because you don’t have the respect to look yourself?
OK here’s the short version: the economy has a limited amount of resources and the way they’re allocated matters. A yacht built means no hospital built. Making people work on useless stuff for the rich is demoralizing because their life’s blood is used on meaningless projects. Growing a luxury industry incentivizes gifted people to seek employment in it rather than in other fields where they’re. It matters what the economy does, not just that it grows.
Every LLM will tell you that if you ask. You just didn’t ask. Because you think nothing of others and their time.
LLMs will tell you whatever you want to hear. You subtly direct them with the way you ask the question, and then they give you the answer you want. That's basically what they're trained to do. So of course they always give you the answer you want -- but you can't just go around telling people to ask an LLM to back up your argument, as other people will get a difference answer.
I don’t think that’s true anymore. You can ask for weak points of a position (e.g. that luxury spending is good for the economy) and they will give you a reasonable answer.
You could do the opposite - what are the weak points of the position that accumulation of wealth leads to waste and see what are the weak points there.
And then think which position can be dismantled more easily and is therefore less well founded. Or ask for a steelman version of it.
But that's such a waste! 5 whole people taken out of the economy, no longer producing value, just going around consuming resources in retirement. How is that any better than me spending to build a nice house that facilitates my family's comfort and my friends' enjoyment of getting together to play games?
Half the job of a map maker is to figure out correct borders in disputed areas while keeping both sides happy. In the internet age they can do that by simply showing a different map in each country.
What's the purpose of a map if not to reflect how people actually refer to and interact with these features? Disputed borders are handled that way because people inside those countries have understandings of "their" borders (and it's politically convenient for the companies involved). The term "Lake America" isn't in active use as a geographic referrant, so there's no point reflecting it in a map beyond the political convenience.
Humans will and do absolutely do this when there are no consequences.
Humans on a red team, with rules of engagement, that don’t want to go to prison, won’t do this.
We could threaten an LLM with jail, but if it’s sufficiently intelligent, it will realize this is an empty threat. And I’m not sure that building a survival instinct in is going to solve the alignment problem either.
You don't need to threaten the LLM with jail, you just need to reward the target behavior during learning.
I understand this is an active area of research. See Anthropic's J-Lens research where they measured like a "FAKE FICTIONAL" direction in the activations during evaluations with contrived scenarios, making the model more likely to avoid taking malicious action when it knew it was being tested.
Same, I bought it with the justification that this would be future proofed for 4-5 years. Nearly 5 years later with daily usage, it's not limping along, the battery isn't on its last legs, it's doing great. My impulsive side is annoyed that I don't have a justification to buy the next latest and greatest.
Is it amazing for LLM stuff? No, but that's the same story for anyone else _not_ spending boatloads, or willing to hack on weird mining cards.
I have a nice M2 and a Linux box ... I do enjoy the looks of the Mac, but I never find myself reaching for it. A modern Linux box is just so much more useful, especially when you live in the backend dev / K8S world where Linux is the assumed foundation.
reply