Yeah, none to the African, South American, or Asian nations who contribute many of the best players though. And they don't share it around their own countries that well either with all the revenue going to the big 5 and the rest can just go sing.
The non-profit makes no profit but it doesn’t mean those involved can’t make profits. An investment in a new jet for the boss is an investment back into the non-profit!
Once you have the plan you don't need to keep the 2hrs of research in the context (which is most of it) you can drop that plan into a file and start fresh for implementation.
That is true, but I feel like sometimes if the conversation contains useful rational it can help to keep it. I think sometimes it is a judgement call, I will sometimes compress the context first.
If I feel like the model and I explored a lot of options I won't want to keep context as it might be confusing.
I think the more you use it the better judge you are of whether you should purge, compress, or just keep the context before executing the plan.
Models work best when they have short instructions and no noise. "Conversation [may] contain" also means "conversation has a lot of noise". It degrades performance and increases cost.
I use handoff skill to ask model to write a prompt for itself.
The applicability of your advice is very model dependent. Some like Claude have very good long context performance, whereas others they fall off much quicker past some threshold.
You're right that it's model dependent, even within Claude models.
I've found Opus 5 far better as a subagent with very limited context window use, which could suggest that it might not have good long context performance unlike its predecessors. (I was one of many tearing my hair out trying to work with Opus 5 for the past week.)
I am not only talking about "long context" performance (context rot), but also about noise that is confusing model about its goal (from correctly extracting operator's intent). I think every model will get confused to a degree, so clearing up irrelevant information from context helps a lot.
> That is true, but I feel like sometimes if the conversation contains useful rational it can help to keep it. I think sometimes it is a judgement call, I will sometimes compress the context first.
Yes, pretty much - if there’s a lot of noise and jumping around and wrong conclusions and corrections, compress and only leave the correct stuff (maybe make some plan file briefly mention what NOT to look at/do). But if it’s all fairly straightforward then can just proceed with the execution.
Most of the time the planning stage ends up short of 200k tokens anyways, it mostly takes hours just cause I’m slow and need to explore the various options - still cheaper than building the wholly wrong thing and having to redo everything.
Compressing the context can also drop important information so it might be better to only do that when you need to / use the plan mechanism/files / do it after completing some large stage of the plan and so on - so a judgement call.
Because I'm impressed as hell with what the labs have done with the data they have, and I really couldn't care less how they got it.
I've already contributed to their training set inadvertently by publishing open-source software and documentation over the years, and as long as they keep delivering the bang-for-the-buck that I'm seeing now with Fable and Sol, they are welcome to keep doing what they've been doing. More power to them.
Corporate news media defending copyright maximalism with their dying breath, on the other hand... not so much. Continued progress in AI (and my continued access to it) is far more important than copyright law.
you may be fine with these companies using your data and selling it back to you but you're also saying its not okay for others to not want to participate. basically people that actually create something of value should be forced to give up that value (not counting your shitty public repo that you call "OSS"). you don't even think they should pay a fair value to train on it, as long as you get create your shitty vibe coded slop, you don't care.
Correct, I don't care. The data they are "selling back to me" has been transformed and amplified beyond recognition in 99.999% of cases.
Copyright is not a natural right. It's something we pulled out of our asses a few hundred years ago to make a few people rich. It can go right back where it came from, if it's going to impede progress for the rest. Don't fight AI, fight for access to it.
asking these trillion dollar companies to pay a tiny portion of their valuation for the data they're using to profit is not "fighting AI".
with your copyright vision do you think your favorite writers or artists would be putting blood sweat and tears into their work if the publisher just steals it and sells it without giving them a dime?
Yes, why wouldn't they? Except for plaintiffs' attorneys, no one will ever ask an LLM to regurgitate today's New York Times or the latest Harry Potter book.
I might someday ask an LLM for the next Harry Potter book, though. That will be where things get interesting, but that's not what's under discussion here.
just because one model is stopped from being released publicly doesnt mean they completely stop. Anthropic has moved on to training the next gen model months ago.
"In a genius move, they hired design agency LoveFrom to handle the exterior and interior execution: that’s headed by former Apple chief design officer, Sir Jonathan Ive."
It’s another 24 carat gold Apple Watch. Makes sense in the design studio, if you have some insane blinkers on when it comes to how people associate with and interact with products in the real world.
I’m not OP but I just want to point out that navigation doesn’t need to mean I am always sharing telemetry with multiple third parties
I have a garmin watch which is great for overland hiking, multiple day expeditions etc
I download the maps and the watch has GPS to plot where I am on that map. My watch doesn’t have an eSIM at all.
Rivian is an adventure brand so if they wanted to design a maps system like that, where I am not continually downloading tiles from open maps or google and sending my location to them and others, they probably could
I just don’t think they have space for those types of features most people don’t care about while they are trying to compete in a rough industry and deliver new vehicles
It's telling just how completely successful the social media revolution has been, when we don't remember that two short decades ago 3rd-party car navigation options that relied on maps loaded on the device and GPS input and that's all. No SIM cards (though they could have done so at the time), no telemetry.
The experience was even comparable to today's experience - I've been auto-routed around a road closure, like, twice in 5 years? And it _failed_ to route me around a road closure probably twice as well?
Why would you be sharing that? There's no reason why the navigation system needs to constantly tell a remote system where you are. Navigation systems don't even need an Internet connection for basic routing.
Navigation through gps means you're connected to a satelite at all times. has nothing to do with an internet connection. You're directly connected to the U.S. government (Space Force)
A GPS receiver is passive, it doesn't send any signal to the satellites. The satellite broadcast their position and what time it is from their point of view, and the receiver computes its position from that.
Also, there are now several countries that sent positioning constellations (obviously to not have to rely on the US for positioning), and most receivers support several: GPS (US), Galileo (Europe), Glonass (Russia), Beidu (China).
GPS and other similar technologies work by passively listening to the satellite signals without having to respond to them in any manner. Location is calculated based on time it took for the signal from difficult beacons to arrive.
