Looks like most/all manufacturing happens in the SEA/China, so I can see the logic that it could be considered a military risk for it to not be manufactured/possibility to scale manufacturing in America.
You should set things like this in Claude's settings.json rather than just asking the model and hoping it will obey. That way it will be enforced by the harness.
As some other comments have said, there are various other options like hooks which are run by the harness and can be used to always enforce certain things (running a formatter for example).
But for your examples, settings are the correct solution. Set an ask permission on git commit and it will always prompt you before committing, and for git commits just set attribution commit to and empty string and it will stop adding its own attribution. https://code.claude.com/docs/en/settings#attribution-setting...
Which government ? And based on the past few month, if your are thinking of the US governemnt; I can assure you that it is actively being harmful to me.
I have no love for SpaceX but at least I can take a subscription or invest and the stock and pretend that those satellites are beneficial to me.
There isn’t a single US government owned satellite that is not actively harmful to me at the moment.
abstract inductive reasoning requires dynamic learning in unfamiliar contexts. All the benchmarks which measure this specifically (in particular, task composition) see LLMs fail catastrophically. There is quite a lot of research published on these limitations now.
>I’ve been seeing this “high-trust society” dog whistle a lot lately
Brushing scams under the dog whistle rug is a cheap shot left wing liberals use to farm pitty and let criminals and scammers get away with it time and time again.
Your white cells are a dog white too, better remove them to not discriminate against bacteria.
Yep, been in the works for a while. Ericsson, Nokia Bell Labs and Qualcomm have been publishing press releases regarding ISAC and telecom providers also sees new potential market of S2aaS, especially if there is a push towards autonomous robots, vehicles that need the data for mapping & training data.
Future networks using millimeter-wave (mmWave) and sub-terahertz (THz) frequencies may collect or infer detailed information about people, devices, bystanders, passive objects, and environments in a sixth-generation (6G) deployment area. It may detect breathing and heart rate of biological bodies.
Some of the smartest people on the planet all in the same room, data at their fingertips… they randomly add it to the training set?
Labs at least must study prompts in an airgapped fashion. From there, consider how they could generate synthetic data to train another model. After, require trusted staff to do multiple levels of independent granular reviews of all fruits of the highest-value stolen inputs. (Or for model training data only, data never has to leave the airgap.)
Definitely risky, anyway. Surely some AI user has sent data, in confidential mode, with a unique shape they expect to be able to recognize if a later model recreated a facsimile even with heavy substitutions… but labs could bring risk of getting caught (over next few years) down quite low with extraordinarily ultraparanoid strategy. (But hopefully everybody is just behaving!)
This does look cool - but I suspect most of the stuff I'd use it with is 2.4 Ghz (i.e. ISM band) - IOT devices, wireless keyboard/mouse, wireless cameras, drones etc
Is a 2.4Ghz version mostly about the larger physical size? Or are there other technical limitations to overcome as well?
And cost wise - would it be 2x this version ($499) - or higher?
Asking this question only a handful of years after a global pandemic...
If the next pandemic is 50% deadly, not being able to make gloves is surely the canary in the coal mine proving we wouldn't be able to make any other PPE.
And no country can rely on another if it's do or die. Other blocs will keep to themselves.
For example I've been using Jujutsu exclusively (as a Git frontend) for years and I don't think about it, I just use it. I reflected on this couple of years. It's existence is completely transparent to me.
I, however, don't agree with sibling commenter that it's a function of time spent with X though. As a counter example: Emacs was my go to editor for 15+ years, last 2 years - because reasons - I was switching between Neovim, Helix, Emacs, Kakoune. 6 months ago I settled with Kakoune.
Even with many years in Emacs, I still tweaked and tuned it. There was always something to do, change, understand. I actively thought about Emacs.
With Kakoune after initial "set me up" phase, it's just as transparent as Jujutsu. Sure, I made complex plugins (for searching, highlighting unbalanced parenthesis and even a GUI wrapper called Kakvide). But the difference is that in Emacs the driver was the tool itself and in Kakoune it's always "I wonder if I can do X".
And so I believe that Kakoune is better tool than Emacs as it's more transparent to me even with a big time difference in usage.
The fun part is that the argument holds just as true for humans that write code - we also run tests to verify our work!
Which is basically what people liking dynamic languages have said all the time - types is only good as long as the overhead they bring doesn't cost more.
I've had to batch-rename files many times over the years. That means:
1. I do it manually over however many minutes. Works if there aren't too many (especially if the pattern is too complex to trivially automate).
2. I make a Python script for it. No way I'm renaming a thousand files by hand.
3. I don't do it. Too much work. The problem lingers forever.
Or these days,
4. I make an AI datacenter eat another town's water supply.
I've never used Emacs. I tried vi(m) nonconsensually and had to google how to exit. A while later, I tried it intentionally and hkjl navigation didn't work because I use a custom keyboard layout, so I never touched it again. Sublime Text and its many cursors for the win!
I'd love a way that isn't miserable to do such a common basic task.
Your parent comment isn’t saying they doubt the assertion, they’re asking for details.
If you say you know for certain, it makes sense to ask how. It makes a big difference if the answer is “I used to work there”, or “I implemented those systems myself”, or “I heard my cousin’s second ex-wife say she heard it from her hairdresser”, or “aliens visited me in my dreams and told me”.
I don’t doubt these companies are lying through their teeth. We have plenty of proof of several cases where they did, to the point believing they are liars is a sensible default, but still I could not say I know for certain of every instance of their lies.
The compute part may be a rack or a cabinet worth of GPUs (though TBH the public designs are currently vague to the point of being artistic impressions), but they also need to come with a PV array big enough to power that, plus a cooling array that's going to be close to 25% as big as the PV array regardless of what unit size they go for in the end.
If they settle on making e.g. 120 kW satellites, that would be about 400 m^2 for the PV and another 100 m^2 for the radiator.
There are various satellite finder apps. I suspect you'll find starlink satellites are mostly too dim to see - with most of what's visible being other older satellites
The -spark variant of GPT was a ton of fun indeed, such a shame it's so dumb though so really hard to rely on. If you were to compare the quality of mimo-v2.5-ultraspeed with anything from OpenAI/Anthropic, where would it be placed ~more or less in your view?