This is proper expectation management.
Companies leaking their v2 plans effectively shut down sales of v1. And if there’s a long time gap to shipping v2, their revenue gets hammered. If a startup, can go out of business. Have seen it happen lots of times.
Humans shopping online allows the seller to offer discounts, upsell services, create serendipity (“How about a lip gloss 50% to go with those shoes?”), and build brand loyalty. Or if you’re a techie, how about 50% off an SD-card with the purchase of a gadget?
This is why retailers create these expensive e-commerce websites instead of just dumping their wares into E-Bay or Amazon. They also do things like web heatmaps and other types of analytics to optimize the UI/UX.
Having an AI agent do the shopping means they lose all that. It’s any wonder they’re going to fight AI shopping agents. Be prepared for a lot more complex captchas when roaming around the web.
Blowing other people’s runny noses, with your mind.
We used to read both versions in high school French class. There was much more slang in French. Many of these were replaced by silly puns in English.
Even the names: Getafix the druid was originally Panoramix. Dogmatix the dog was Idéfix (this is actually a pretty good translation, keeping the core idea of single-mindedness, plus it has Dog in it).
The chief and bard names are the worse. Abraracourcix is a reference to someone prone to violence in French, which is why he keeps getting angry and red-faced. There’s a whole plotline about him having to go to a spa so he can lose weight and relax. Not sure why they renamed him to Vitalstatistix in English.
And the noisy bard goes from Assurancetourix (comprehensive insurance joke) to an unsubtle Cacofonix. But to answer your question, most of the bad puns were added in the English translation.
FWIW, they did a reverse butcher job with Harry Potter books. The French versions literally translated the British expressions word-for-word to the point they made no sense.
Pretty cool.
I’ve been noodling on what else you can run inside a GPU shader. Problem was data persistence. This is a pretty interesting approach.
Wonder what kind of performance you get running something like this on a bare, high-end GPU (outside the VRChat/Unity scaffolding).
Only 6 takers on the KS with a long way to reach funding level. My guess is they’ll have to relaunch while focusing on end-user benefits and software instead of hardware specs.
The Qualcomm IPQ9574 is a pretty high-end WiFi 7 platform. If someone already has a decent router, it’s a bit of a waste. The main processor is the Rockchip RK3588 module which already has dual NPUs. Adding another AI coprocessor means the system drivers have to be tweaked properly to use the right coprocessor. That’s why I think it’s important to see how they’ve implemented the software.
The RK3588 is a couple years old. Rockchip already announced the RK3688 but my past experience with them was they first released mobile versions and it took a while (1-2 years) before they made dev boards and server BSPs available. The 3688 also has a much better, faster NPU w 20 TOPs – not as beefy as the Metis with 214 TOPs – but OK for basic local inference.
All the communication slots are good for remote office or High Availability – a bit wasted IMO for home use, unless you need LoRaWAN, satellite, or multiple 5G lines.
If you badly want to use a 3588, BananaPi makes a pro board at a fraction of the price. Otherwise, for basic home server use, an old Intel laptop or headless desktop, reflashed with Debian or Ubuntu will do.
Again, I really like the hardware mix they have. It’s great for a small office or a research lab, but IMO a bit overkill for home use.
Pretty loaded package. Love how expandable it is.
In its current configuration, it looks a bit underpowered for local AI (16GB), but there are a lot of slots to add your own M.2 board.
I’d be concerned about driver support. Given all the hardware, the burden is on the admin software to help with configuring all the knobs and buttons. Didn’t see any mention of that on the KS page. Mostly hardware specs (which again, are pretty good).
I’d want to know more about the software. If it requires downloading tar archives and manually configuring things, it’s not meant to be used by non-devs. They claim it was designed for medical office use, but for that they would definitely need an end-user friendly interface with a LOT of sane defaults.
Also, allowing only 15 days for a hardware KS is a bit strange. It takes a while to spread the word in the device community and get backers. Not much time to make a decision or get budget approved for a $2K+ device.
Side effects: sharp fangs and uncontrollable drooling.
I sat hunched under my elderly mother’s car, video running on the phone in one hand, showing how to remove a ridiculously complicated headlamp cover, just to replace a busted headlight. Hand scraped and bleeding, but it got done. Would have NEVER figured it out otherwise.
Thanks that one YouTuber.
This, in a folding, commuter e-bike.
If you’re in the Microsoft, Apple, or Google OS ecosystem, or use one of the appstores, it’s likely they already have this age information, or could verify it via indirect means (credit card checks, public records, etc).
If you have set up a family plan, then they already know the age of the children to enable access to appstores.
What the OS will do is likely store some sort of validated token or system level cookie that says “adult = true” and we’ll never even hear about it. My guess is Unix-based distros will do it too and avoid having to piss-off customers by implementing third-party attestation, or worse, per website.
The reason the European systems had to ask for a third-party verification system with you having to upload IDs and all that was exactly because the OS didn’t do this. Now, it’s going to be done once (and not require any user interaction) and that will be that.
Personally, if implemented properly, I think it’s a much better system than having them upload private data to some nebulous third party. That’s a big IF. They could cock it up and make everyone upload face scans and pictures of IDs. In which case we’ll be back to rubber masks and stupid override games we’ve seen.
The BBC Microbit was designed for exactly this scenario: https://microbit.org/
If you want them to have a more desktop-like environment, the Raspberry Pi has kid-specific projects: https://www.raspberrypi.org/learn/
Or you can get a Pi Kano kit and it has everything you need: https://www.kano.me/
Privacy, shmivacy.
If someone gets physical access, all bets are off. Just wire a ChipWhisperer (https://www.newae.com/chipwhisperer) between the TEE and the CPU or GPU, and it’s trivial to get around all protections. Throw in clock or voltage-glitching and…
Can someone plausibly explain the Bending Spoons business model?
Big chunk of debt financing for decrepit brands with old tech. Other than strip-mining user data, I can’t figure it out.
Was in San Francisco last Sunday. Ridiculous number of robotaxis everywhere, most of them empty. Mostly Waymos, but also a few Zoox ones and 3-4 others with a training driver.
I kept trying to get away from them, but there were groups of 2-3 driving around, boxing me in. Mostly clustered around popular tourist spots.
On top of that, San Francisco has now started implementing traffic cameras that will snap your license and automatically send you a speeding ticket if you’re 5 miles over the speed limit. I get that it’s a congestion management thing, but add that to all the robotaxis and I’m not sure many people will be driving in San Francisco.
Come to think of it, that’s actually a good thing. Carry on.
That’s pretty decent range, and having solar panels trickle charge all day means for a lot of people it’s basically at zero fuel cost. That right there is the big selling point.
If they figure out how to lop off a foot or two from the back without destroying aerodynamics and solar charging, it would look and park more like a normal vehicle.
Gullwing doors are fun, but a bigger person might have trouble fitting through that gap.
FWIW, Toyota experimented with a three-wheel, two-seater a while back, but range was nothing like this: https://theautofuture.com/2014/07/12/toyotas-three-wheeled-road-ev/
The power company AI will generate the regulatory reports. The regulatory commission’s AI will summarize and score them for faster turnaround. Only 30% chance of hallucination making it through each cycle.
After a few iterations, they will get fully approved to deploy the giant hamster wheel with Bluetooth, powered by Nuclear-flavored RedBull.