I paid around that for a 5x10 unit in a medium sized midwestern city for a year during covid when I worked fully remote and moved back to my parents' house for a year.
For general purpose use this is interesting, but if I'm just using an LLM for coding, does this matter at all? I would hope something like a new java version after a model's publish date can be handled and understood by the model through tool calls and context even if it's not explicitly in the training data, the same way the LLM doesn't have my existing code or the plan to change it baked in from training.
The inference time results are quite different. These models haven’t just been trained with tons of docs, blog entries and videos about Java 8. They’ve also been trained on tons of code examples from simple to complex real world ones.
You’re not going to get all that just by pointing the LLM to the recently released Java 27 documentation. That information is also potentially adding tons of content to your context, which is already filled with tons of other data (your code, other recently released libraries it has to get documentation for, etc).
My experience with Claude is that it doesn’t handle its own training cutoff properly. It responds as though its cutoff is today. Gemini is even worse about this.
I’m going to add this site to my agents files so that they’re explicitly aware of their own limitations.
If someone launched a nuke in space how would other nuclear powers react? I doubt they would assume "this is probably just an EMP and not a direct nuclear strike on our territory"
If you had time to plan it, you'd probably want to put the warhead on a satellite launched by a normal satellite launcher. It'll look benign until you detonate it.
I don't know how someone like me would even tell if this treaty is taken seriously. For example, going to space is really hard and has historically been somewhat unreliable at the best of times, which may mean that back in 1967 everyone was thinking "we don't even want a 1% chance that someone's failed launch to drop a 99%-usable warhead on a random location where some tiny country or organised crime group can find it".
Not really relevant for public service announcements like gp is talking about. I just want a live feed of my local services updates. I get weather alerts from my city and county over text (contracted by the government through a third party platform), I would take them as a dedicated app or rss feed or anything. I'm not connecting to my friends or family through a weather alert or train delay notice
I find texting pretty decent but not ideal. Most alert systems offer it already, you know it's going to be a short message, and you'll be informed quickly of things instead of having to reach out somewhere.
Maybe there should be a texting app where you can make pages of contacts to swipe between, and you can customize notifications. If you want to get fancy, maybe local AI to determine if something looks important or not? They'll always have ALERT or something in the message.
By killing Google Reader they essentially killed RSS. Yes, I know there are other clients some of us still use, but Google's was the biggest and most popular.
I don't think I agree with this characterization. If you consider the open standard to be indistinguishable from the proprietary client, I'd argue it was already dead before the client got killed off.
The problem here is you need a trustworthy government for nationalizing to make a difference. The current US admin started with DOGE and a crypto rug pull.
Waterfall means designing and building everything before rolling it out to users. Researching a problem before you start tackling it is just general good advice. You can be more effective while moving fast and breaking things while still taking some time to know where to move and what is ok vs not ok to break.
It's crazy it is resold but we don't even get it for our own use. Budgeting with credit card statements would instantly become so much more useful with less effort.
I don't think it's JUST Teslas, any car with its own cellular link is a total non starter for me after that Jeep article 10ish years ago. Give me a key fob, carplay for my music and maps, and that's all the smarts I need. Keep the heated seats.
Our customers have spoken and we are listening. The 12V power outlet in your car now requires a $9.99/month unlimited use fee or an affordable per-watt charge you can pay with a CarBucks card found in all major retailers!
In the longstanding tradition of leading on the forefront safety, we introduce Engine Check coming up after mysterious, unaccounted for, unsigned voltage drops.
Can't be too safe if it could be a short, right?
"Tinkerer mode" only £14.75 per month*
*additional checks during the annual service apply.
Even Shingy called chatgpt "chat" when he went on Ed Zitron's show. It was very weird to hear from a guy who has been around tech world for such a long time even if he's non technical.
I'm a pretty seasoned software dev, but I've also kept the frontier AI race at arms length from myself. I run Claude alongside my day to day workflow but don't really get overly concerned about the latest and greatest models, as I'm not leaning on it to do my job for me I suppose.
In my mind I'm after a stable tool, not wanting to dabble in the bleeding edge. I did that back in the early days of web coding and have settled into a much more subdued kind of industry/life in that regard.
Everyone calls it chat. I am technical and even I’ve come around to calling it that because that is what everyone, including very technical people, call the product.
reply