• 0 Posts
  • 42 Comments
Joined 1 year ago
cake
Cake day: June 15th, 2023

help-circle

  • I don’t know about this specific case, but it’s common for the big name researchers not to do any actual research or play any direct part in generating their images. That’s often done by kids - 25 year old grad students, even 20 year old undergrads - or other trainees. Those people may not appreciate how easy it is to detect image manipulation and are still learning what kinds of ‘refining’ of imagery and datasets is acceptable, while the PI that pays their stipend or sponsors their visa rages at their inability to get an expected outcome or replicate a previous result.

    Not saying there aren’t people out there just flat-out frauding, but these are group projects with a structure of trust and pressure that can muddy assignment of culpability. Like any committee or corporate action, it can be tough to say that any one individual is the guilty party or which people where just going along with the group.





  • The Android app should still be fine. I’d expect Apple’s move to be followed by a lot of creators adding a “Don’t use the iOS Patreon app” to their profiles.

    I mean, apps that are just the website are a bad idea in the first place, but this specific problem is entirely contained to the iOS app. If some people prefer an app to a bookmark, that’s on them.



  • Once you have a microcontroller running things, adding new features is just a matter of software. Doesn’t add to the BOM, doesn’t complication production in any way. There’s almost no marginal cost to techify everything, and the people who don’t want those features can just not use them. The small minority of people who want a repairable car that they can understand and maintain in their own garage are undesirable customers who reduce after-market revenue.


  • tburkhol@lemmy.worldtoTechnology@lemmy.world*Permanently Deleted*
    link
    fedilink
    English
    arrow-up
    10
    arrow-down
    3
    ·
    2 months ago

    In the US, plug-in hybrid is a decent way to cover the breadth of consumer desires. Get a battery big enough for 50 miles of daily commuting, but have the ICE for 500 mile holiday trips. More complicated, having both power systems, and you still have the tie to gasoline, but you don’t have to lug a massively oversized battery pack everywhere you go and you still get most of your transportation energy from the electric grid.




  • I don’t so much care where it’s made. The real selling point, to me, for Pi is that their products are well documented, in English, and solutions for problems are easily googled. There’s tons of SBCs out there, some of them even inexpensive, but I can’t tell if any are going to last longer than a single production run. Meanwhile, I can still buy a Pi 3 after almost a decade. Or I can take the hat I made for a Pi3, plug it straight into a new Pi Zero, and expect it to work without changes.

    IPO is a big step down the path to enshittification, especially when there’s no clear, dominant alternative.


  • Even if you ignore all the neuromodulatory chemistry, much of the interesting processing happens at sub-threshold depolarizations, depending on millisecond-scale coincidence detection from synapses distributed through an enormous, and slow-conducting dendritic network. The simple electrical signal transmission model, where an input neuron causes reliable spiking in an output neuron, comes from skeletal muscle, which served as the model for synaptic transmission for decades, just because it was a lot easier to study than actual inter-neural synapses.

    But even that doesn’t matter if we can’t map the inter-neuronal connections, and so far that’s only been done for the 300 neurons of the c elegans ganglia (i.e., not even a ‘real’ brain), after a decade of work. Nowhere close to mapping the neuroscientists’ favorite model, aplysia, which only has 20,000 neurons. Maybe statistics will wash out some of those details by the time you get to humans 10^11 neuron systems, but considering how badly current network models are for predicting even simple behaviors, I’m going to say more details matter than we will discover any time soon.



  • I think about the Vision like I think about a new Gucci bag or a new set of Air Jordans. There’s a small, but very visible, community that is super into that product, probably for reasons not related to its actual functionality. The difference is that there’s a lot of overlap between Apple fans and broader technology enthusiast groups, where we’re more isolated from the Gucci and Jordan communities. There are lots of brand-based fan groups who will happily accept branded merch or content, but not interpret that as ‘advertising.’

    The rest of the world tolerates spyware and especially ads if they feel like the product is worth the intrusion. There’s a reason Meta doesn’t have a logo watermark foating in the corner of Quest view field. There’s a reason VR is still very niche, almost entirely limited to gaming.

    Maybe Vision’s AR experience will change that. Maybe viewing your entire life through a video camera with overlaid graphics has real-world value beyond privacy in co-working spaces. I doubt that value is $3000 and think Vision is more like Apple’s Newton than Apple’s iPhone.


  • I would rather spend (modestly) more time checking my own than less time standing idly with nothing to do but watch some kid checking out my goods. It feels better to be an active participant. Where it breaks down for me and my 12 items is when all the self-check lanes are clogged with people trying to ring up a full cart of groceries, who still haven’t figured out how to work self-checks, who are encumbered by a baby in one arm and a phone in the other hand, or who just can’t move all that well.

    Managers using the presence of self-check as an excuse to understaff the actual checkouts makes all of those problems worse, and makes the checkout process suck for everyone.



  • Some of that turn is physical plant. Kitchens, especially, are built to serve human forms, where tech solutions to food prep would rather be stand-alone boxes. It’s a far harder problem to make a robot that uses a restaurant’s existing grills, ovens, and deep fryers than it is to make a box that turns out perfect french fries. It’s a riskier proposal for a restaurant to replace its fry station, where a human can make fries, onion rings, egg rolls, or whatever new fad hits tiktok, with a fries-and-rings-only box with less than 10 years commercial proof. Generative AI, for all its faults, is just code that runs on a computer you already have, or maybe in a cloud service with zero physical footprint. Relative to replacing your barista with a vending machine, trying ChatGPT for a quarter or two is practically zero risk.