One of my kids is a huge fan of Dog Man by Dav Pilkey. This particular page from The Scarlet Shedder really captures the mood today.
TechTakes
Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.
This is not debate club. Unless it’s amusing debate.
For actually-good tech, you want our NotAwfulTech community
Scam investigator Coffeezilla is working on the Rabbit AI guy, and he's off to a good start. Most of the other press is about the device not working, while this focuses more on the scammy NFT / crypto past of the founder.
somehow missed this previously, but: Redhat Chatbot Linux!
Today, Red Hat announced a developer preview of Red Hat Enterprise Linux AI (RHEL AI), a foundation model platform to seamlessly develop, test and run best-of-breed, open source Granite generative AI models to power enterprise applications. RHEL AI is based on the InstructLab open source project and combines open source-licensed Granite large language models from IBM Research and InstructLab model alignment tools, based on the LAB (Large-scale Alignment for chatBots) methodology, in an optimized, bootable RHEL image to simplify server deployments.
the "LAB methodology". I wonder if someone broke out the rack and thumbscrews to get a name that tortured
given what I've seen people do to ubuntu-based machines to make their kerases and tensorflows run, combined with redhat's historical trend of extremely bullshit repo pains, I have to wonder what nightmare experience this will result in
So who is going to be first to set up a grifty EA think tank where the thinking is all done by LLMs, and why shouldn't it be us.
Bold of you to think we have not already started doing this to mess with the EA/LW minds. Why do you think we suddenly talk less about all the outrageous weird shit they say on LW? We don't want to out ourselves. For example
Pull a Rabbit AI and use Playwright to operate an Emacs instance running Eliza. Say that buying more infrastructure will yield better results. Pocket the money that was for supposed infrastructure investment. Profit!
Kevin Beaumont got access to Copilot+ and hoo boy it’s even stupider than it sounds:
I got ahold of the Copilot+ software.
Recall uses a bunch of services themed CAP - Core AI Platform. Enabled by default.
It spits constant screenshots (the product brands then “snapshots”, but they’re hooked screenshots) into the current user’s AppData as part of image storage.
The NPU processes them and extracts text, into a database file.
The database is SQLite, and you can access it as the user including programmatically. It 100% does not need physical access and can be stolen.
and the anti-backlash begins, because snapdragons are magical and microsoft has never fucked over its users
you know what really fucking gets me about this thing? even "before" all of the security and DV etc arguments (and, really, they are all valid)
consider the average consumer-human (present version of) windows desktop/laptop/workstation that has survived for more than a year or so without having been re-imaged. those are already often an unusable nightmare of built-up cruft in sticky corners
and now there's going to be something that'll do constant capture and analysis every fucking moment it's running?
how the fuck is this not going to go wrong? imagine the first time that sqlite table shits the bed. a dodgy character format goes into it, or a bit of disk corruption hits, or a file locking (lololol) issue hits, or .....
I just ..... I just really can't imagine any of this is going to go very well at all in the real world
I’m calling it, when this bullshit gets deployed widely a bunch of absolute shitheads are going to dismiss Kevin’s post with “that was the prototype, they changed everything for the real version” and the only thing that’ll change is they’ll superficially encrypt the SQLite file
pbccyhf_erpnyy_cevi.db
love how the MS shill eventually reaches:
Okay, here's a question; when has the casual consumer ever cared about privacy?
every fucking time! and the answer seems to be: they care enough that privacy’s a core part of Apple’s marketing, but microsoft’s core competency is still managing Maslow’s hierarchy of needs enough that their users are on average just barely more inconvenienced by switching operating systems than they are dealing with whatever new shit’s being sprung on them (and this latest new shit’s gone from annoying to actively dangerous, so the tactic is to convince users it’s irrelevant to them until it’s made mandatory, as Microsoft does, and then the pain of suddenly switching operating systems will be even greater)
27 years of trying to make clippy mandatory and they’re still fucking at it but surely this time they’ll listen to their users
the fuck? any of you happened to have seen this?
I don’t know if I’ll get it on my one droid (both because I’ve turned near every setting and tracker and whatnot off in advance, and because ZA - less likely to get shit like that in advance) but loooool
they’re all just shoving this shit into every damn input control
coming soon: mousegpt?
coming soon: mousegpt?
unfortunately, logitech’s ahead of you on that one, and of course it’s the type of lazy shit you could design in an afternoon, so that means they’re gonna sell you a $50 AI Edition mouse:
Logitech Signature AI Edition mouse: $49 @ Logitech
This is a new version of the Logitech M750 wireless mouse that comes with a new teal button on top specifically intended to be used for the AI Prompt Builder. Of course, you could likely remap it using Logitech's software to do whatever you want.
and I realize the article I linked is essentially a paid undisclosed advertisement, but holy fuck:
I just tested Logitech's new shortcut to ChatGPT — and it's a big time-saver
Get ready to summon ChatGPT with the press of a button
it’s a fucking macro key that opens a popover that pastes text into ChatGPT running in a browser get the fuck over yourself
fuck me this is why I make peripherals from components these days instead of buying keyboards and mice fucking laden with this bullshit. when my current (no-name) vertical mouse kicks the bucket I’m gonna have to see what repairable mouse kits are available to replace it
Thank god I can have a button on my mouse to open ChatGPT in Windows. It was so hard to open it with only the button in the taskbar, the start menu entry, the toolbar button in every piece of Microsoft software, the auto-completion in browser text fields, the website, the mobile app, the chatbot in Microsoft's search engine, and the button on the keyboard.
when my current (no-name) vertical mouse kicks the bucket I’m gonna have to see what repairable mouse kits are available to replace it
If you like trackballs, Ploopy is great. They’re open source all the way down, mostly 3D-printable, and actually better than most commercially available trackballs.
They also have a mouse kit, but I can’t speak for that.
they’ve been high on my list to check out! last I checked I don’t think they have a pure kit version, but if the quality’s good then I don’t mind paying for a built version as long as I can replace most of its parts when the time comes
not much of a sneer but a passing chuckle:
Received: from <snip>.net (unknown [snip])
Subject: ***SPAM*** Binance Distribution of MyEtherWallet (MEW) Airdrop
my spamtrap addresses get these every now and then. I'm still amazed that I get that type of subject line in tyol 2024 - spammers are usually far more agile
Not a sneer, and as such NSFW, but here is a short hbomberguy style video: I Debunked Evolutionary Psychology, featuring friend of the sneer PrimalPoly, by the talented münecat.
A well-respected terminal emulator for the Mac decides to slather on AI sauce, and most users are not happy. But there are others who are outraged that slathering AI sauce onto a product is even seen as a negative. Let's take a break from HN to take a look at the tortured souls on lobste.rs who won't let this aggression stand, man:
https://lobste.rs/s/7aglnr/psa_iterm2_now_has_chatgpt_integration
archive link in case mods decide to nuke it
Personal favs:
This is the sort of hatred and resentment that is uncomfortable to see on a public forum.
Hi, this is the internet, you must be new here.
I’m mostly checked out of here as a result until the mods get it under control.
Don't let the door hit you on the way out.
Putting the "evangelist" in product evangelist.
thanks for mentioning this, I'm definitely going to have to nack that update and send a gfy mail
query mail sent
hell, if anyone else is curious to also poke:
I presume I'm not the first nor (will I be the last) to ask about this,
but for the weight of the argument it's still worth asking
are you planning on keeping the openai garbage directly integrated in
iterm2? and "just turn it off with a setting" isn't a good answer tbh
I don't want it in the terminal I use. I don't want it in any software I
touch. openai are a deceptive abuse company led by an abusive liar, and
it saddens me to see yet another good thing get taken up in the hype
these charlatans drive
no "I've been $x for $y" pleading arguments here, I don't want to try
bend your ear with any number of platitudes and such you've heard from
others - just straight-up asking the above
Noooooooooooooo! Argh, I'll have to seriously consider using the fork FML.
EDIT: Not strictly required since apparently you have to provide an API key for it to be enabled, still it's not encouraging that the main developer thought this would be a good idea.
You've got to love the prompt jank: https://github.com/gnachman/iTerm2/commit/755dc2ed881d853f495ffaea2498452915e5e8cd?diff=split&w=0
EDIT 2: Given direct access to bad AI code to a dev workstation is bad enough, but given that the console is a primary way to connect to servers, where more havoc could be wrought, this is terrifying, I mean sure devs were already capable of bricking enviromnents, but supercharging "knowing just enough to be dangerous" is NOT a good idea.
Saw this update yesterday and it pissed me really off. I've been using iTerm for almost 20 years at this point and it's a really good and well-integrated terminal application for Mac OS, and no one asked for it to add spicy autocomplete.
Or maybe I'll just move over to Alacritty. I use that one my Linux machines and it's actually good, but the attitude of some of the developers when it comes to feature requests and contributions really makes me like it less than I could.
I don’t like AI hype either but I don’t feel the need to use Lobsters to rally groups of people to mass punish projects who add optional AI-friendly features. That is quite disturbing.
ok. I tried writing something sarcastic and sneery but alas, I don't think I can mock this person hard enough to satisfy my ethical obligation.
Smearing the developer as “ai-brained” is hateful.
wake up babe new slur just dropped
Soon they will discover that they can use AI to autogenerate new minority definitions for them and we on the left will be powerless to stop them.
(We will mock them for it, and tell them this new made up thing doesn't count, but we cannot stop them).
tbh the enforced civility and de facto ban on social issue discussion on lobste is not helping. (but after last “polite discussion” i had with the moderator i'm not going to bring it up anymore, just suspend the account or wait to be banned.)
yeah frankly that sort of shit is why I haven't even tried to get a lobsters account
bullshit should be called out as bullshit at any point in time. dressing it up in niceties is playing the wrong game, and has been abused by the world people around for years