It's not just shooting at static monsters in an empty room, this will go after monsters, make decisions about prioritizing health, changing goals, and finish the level, etc. It's not unique, it's just more impressive than the demo they're using.
I noticed Trader Joes was selling champagne grapes the other day. Things must be really, exceptionally grim. France has been stockpiling wine in state owned warehouses (to keep it off the market) and paying farmers to tear up their vineyards.
If you're looking for a more impressive doom example, I put together laya-duum which uses open source micropython implementation of doom (duum) and freeware freedoom1.wad. It uses the standard jev api and will play through the first two levels to completion: https://github.com/Hadlock/laya-duum
Selling oil out of the SPR is the new One Weird Trick Military Planners Hate! Biden was a generally fine president, but pulling oil out of the SPR to balance the budget was an extremely ugly way to do it, and hopefully that loophole gets plugged going forward.
Xiaomi already showed that you can train a better-than-Sonnet level model in 8 days for about $850,000. Anthropic really could release a new sonnet version every week if that played to their marketing strategy, but nobody cares about sonnet so it gets released on slow news days to fill space.
S3 cost growth is so gradual most businesses will just absorb the cost. We have some physical backups, but everyone is more than happy to not be shuffling around physical hard drives, and pay Amazon to deal with that and securely store it.
I finally stopped using (S)FTP when i realized that all the ftp clients now have native support for S3. It turns out if you have a business partner who wants to use sftp, 99.9% of the time of they upgrade their client, the upgrade has S3 support, and then you can just use modern tooling on your side. And when they finally automate, they can also use modern tooling.
I recall having access to an Apple II in first grade to play Oregon Trail, and later the full color Macintosh LC II with a mouse in 2nd grade. I think besides using a word processor once a week in computer lab, and typing class in middle school, there was little to no value in computers in school. I was a heavy user of Encarta in the library during rainy day recess. Giving every kid a laptop or tablet seems like a financial hand out to whatever company is selling them.
>they're trying to create that "System One" narrative
I think you answered your own question. Executives are going to ask two questions, 1) how is this different/why does it matter and 2) how will i use it to make money?
Leya came to market more than a year before Jev, and failed because nobody understood how to use it, and he was unable to market it properly. Jev used this strategy and did not fail.
I've run into hitting limits on the personal plan perhaps twice since the beginning of the year. But also I don't use the personal plan for coding tasks between 7am-noon M-F.
The Jev alternative, laya is only 334m parameters and you can post train it as the workflow is open source. The downside is that laya is only 1024 tokens, and Jev is ~32,000 tokens. I think for many tasks 32k tokens might be too small. You could probably design a task complexity router with something like qwen 0.8b with similar speed and a much larger context window (260k token). As always, "it depends". 0.8B is almost as fast as Jev without any of the limitations.
Not having to faff about with Meta's developer program or whatever to side load my own homebrew stuff, and then get locked out, then have to do whatever hoop jumping six months later, is a huge bonus. I have an OG Quest 1 and the damn thing needs a new copy of meta quest ...whatever, installed every 2 weeks if I want to use it. It's exhausting.
I stopped using signal when they made their weird change about not supporting SMS due to... whatever weird problem they had with normies. Come on dude. Even my realtor was on Signal. Instantly killed the product.
I deeply regret buying my LG. Originally the plan was to never connect it to wifi, and use an external device (chromecast, apple tv, or whatever) but my wife was too excited to try out whatever features. Unfortunately we can't sell the TV for anywhere what we paid for it new (to replace it with a different brand) so we're stuck with it for another ~7-10 years, or whatever the standard TV replacement interval is these days. New "apps" keep showing up on the main screen, which I have to "delete" from the main screen once a month or so (they're not deleted, they're just undeletable apps in some massive library of preinstalled spyware we can fortunately mostly hide). Apparently Sony OLED screens are LG "glass" but without the spyware.
Inflation helped me kick my junk food and fast food habits. I still buy groceries of course but when prices rise, luxuries are almost always the first items to go.
I've been training different architecture 1.33b and 3.33b models using nvidia megatron for about a month. You can train a 1.33b on a 46gb mac and 3.33b is about as big as you can go with a single 96gb blackwell with regular checkpoints etc
You can warm cache regularly used engram/n-gram if you're willing to merge PRs into a personal branch and build it yourself. I was trying this with qwen 3.8 flash next and the n-gram to get it to fit on my very average gaming desktop (it worked)
Judiasm is monotheistic, Christianity, if you ask anyone besides a christian, is polytheistic, you have The Father (1), The Son (Jesus is also God, or A God, 2), and The Holy Ghost (3). There was a big, big brouhaha about this, the Nicene Creed is a core part of protestant religions to help remind everyone that they're technically a monotheist religion and somehow Jesus is a Flavor of God and not separate despite most of the new testament has him as a fully separate person with his own separate agenda. Even if you discount The Father as a legal entity only you still end up with a polytheistic religion. Writing Jesus into the canon 2000 years later as a seperate but equal entity was bound to be fraught with continuity issues.
MCP is great for attaching chat interfaces with it. It's so good it's slightly detrimental, people start trying to build dashboards with it only to find out you can't make deterministic dashboards with a non-deterministic workflow. The number of people who want API access, now that they've been properly exposed to the data, has doubled month over month since we put in the MCP back in April.
Rather than teach the agent where grafana is, how to use a sentry trace id to find a otel traceparent, where my ALB is, I'm what clusters do what, which namespace prod is in, what our stack looks like, that thing that looks broken actually isn't, etc etc etc. I just paste in the sentry trace id, a guess of what might be wrong ("i think we overtuned gunicorn again" or "developer bob pushed short sha 123456 and nothing is working") and say "use your investigate skill" then get up and go get a coffee and usually by the time i get back i have an investigate doc filled out from a common template in my notes repo. The agent almost never spends any time spinning it's wheels finding out what the various environments do, where they're located, what our metrics and logging looks like, how to access it etc etc.
To be fair it's the only skill i have/use but I got real tired of explaining the same 12 things over and over. Having it document every incident means I have a dense library of every problem we've run into over the last six months which helps identify recurring problems for RCA
The japanese branded car outselling the domestic branded version in 2025 shouldn't surprise any millennial. If you grew up in the 80s or 90s getting shuttled around in a GM you almost certainly remember a piece of the interior coming off in your hand, or if you're an elder millennial, writing your name with your fingernail on the UV damaged interior of a late 70s/early 80s GM car that's only 3-5 years old. It's hard to shake that kind of shoddy build quality even decades later.
Meanwhile my friends kids are turning 16 and inheriting the honda accord with 250,000 miles I remember getting shuttled to and from soccer practice when I was 12 in the 90s
There's definitely a wide variance in laundry. Laundry is something that I can do to kill time while boiling pasta on the stove, for my partner it is an entire afternoon task.
>The key is you're doing it yourself, so it takes your time away from something else.
There is still a cost to the business owner, they have to find a place to hire a freelancer, interview and then explain what problem they need solved, go through several iterations with them (either via meeting or email, or both) etc etc. If you're a small business that's not setting up an online storefront, the time spent setting it up and building it yourself is possibly less time. And then there's the bigger headache: what happens when the freelancer stops returning your calls? For the average small business website/page it is 2-10 pages. The average business owner can probably crank out that site with AI in less time than setting up a single interview.