Posts by Sampler 1174 publicly visible posts • joined 16 May 2007
Sadly my boss likes me to use Claude too, but I can't use Claude within Zapier to adjust Zaps, and I can't use Zapier's AI as it just doesn't work and is pure garbage.
Hoping for the day it goes away and I'm free to just actually do stuff and not provided documentation to show I'm using the AI to achieve time savings that simply don't exist..
We have this new super secret AI that we've shared with only the companies who are highly invested in the bubble and they say it's amazing and better than anything we let you plebs use, super honest trust me bro...
Meanwhile I'm here giving the Zapier AI a go and it's misunderstanding requirements, ignoring others, inventing variable names to pass between modules that don't exist and then telling me it's made updates whilst the UI's sat there showing absolutely no change, repeatedly...
Can this bubble just crash already? I want to buy some hard drives, you know how hard it is to come across hobo's with decent value kidneys these days?
Several seconds sounds terrible for proprietary - I use an app on my old tablet and can use it as a remote screen (USB or wifi) and the performance is largely the same between the two.
I've never tried to watch media or do anything responsive over it, but browsing is perfectly cromulent and keeping teams/outlook on it when I'm out with my laptop so I have periphery of new queries is handy. Really can't see much of a delay at all..
Didn't like the computer they got me so gave it to a colleague and use my own I built out of some spare parts, I also run my own dns server with an encrypted connection between my machine and it so very comfortable to know my key presses or internet activity aren't being tracked for "productivity" and I can safely read el'reg in work hours.
They do it old school here, do you do your job? Yeah, great, crack on then..
Not doing as much as you should be, we have a chat about where you need training or why it doesn't get done as quickly as we think it should and how to solve that, you know, managers actually supporting the people they manage, I guess it's a bit old fashioned, but it works for us..
Another article about how someone sorta, nearly almost made something with AI whilst spending considerable money whilst holding it's hand...
Can we just stop with the propaganda for the hallucination grift as a service platform and just let this bubble die already, I want to buy some hardware...
They used to do facial recognition on all images, that's got to be harder to say this face is same as this face in a different orientation and lighting than the faces on record, we're talking matching a similar enough image.
We know this as they used to prompt you to tag your friend in the image, so, yeah, I think they have the processing capacity. Plus, Immich is doing the (in my view harder) face stuff for me on an N100 chip so it can't be that intensive..
I think this would be a job for AI recognition at least, for actual implementation, like, once an image is flagged for it to not be uploaded again, like, if I crop, flip, change the colour depth or however they flag the image (like, if it's a basic crc then it's trivial to change a file hash without affecting the image).
So, as much as I hate AI, having some automated system that can, we a degree of confidence, say these two images are the same, this one's just been arbitrarily adjusted, would actually be a good thing.
(appreciate that's not what you were talking about, but started a thought)
4x8TB's in a RAIDZ5 array in my nas, 22TB external hot copy connected and 2x12TB's in a RAIDZ1 on an offsite nas.
One of the 12's needs replacing (ideally both to increase the storage volume to get closer to the primary) and I ordered a Jonsbo N6 back end of last year and it just showed up this week, now I can't afford to put fresh disks in it.
I get AI needs storage (and can't wait for that bubble to burst), but surely these are large disks, which scalers are buying the 8-12tb range, why can't I have my enthusiast drives?
Reasonable?
How many regular, everyday android users, do you feel are not using the Play Store outside of China where it's not allowed?
Most don't use third party stores, fewer still will be looking for apps outside what's also available on the Play Store and therefore already signed up.
The amount of people this "security" feature is protecting is infinitesimally small and, in all likelihood, using these alternative stores for good reason and because they're technically savvy enough to know the risks.
Given the premise falls short of reality, there's clearly another reason that Google are doing this, one they don't want to admit publicly, and that should be a very big red flag.
Exactly what I thought - I'd even put this in a bug report, the agent was reading unrelated files and wasting tokens, limiting the products use within the finite windows of opportunity (as they split the daily allowance over four hour periods meaning you can't use two thirds of your allowance during a standard work day..).
Have since cancelled Claude as it's no longer worth using, it was useful at one stage, but they kept tweaking it making it hallucinate more and take longer to complete a task with more refinement required that instead of getting five or six queries done a in a window we were going around in circles on one.
Not that any replacement has been as good as this used to be, which obviously we're not allowed to go back to the earlier, actually productive models... which I had the vram to run this locally and then not care about the rapid enshittification.
What I don't get every time I see one of these articles; AI barely made something not really functional in the real world, but if we squint a bit and forget several things it's kind of amazing it nearly did something..
I guess it's in belief that it will get better and someday work, but, it's chaos inherit in the system, they work on probability from their training data with a few things tacked on, calling them AI is laughable as there is no intelligence, just chaos iterated until it's close to being right, but never knowing what right actually is.
Can't wait for this to be realised, that AI will always be "nearly useful" but never cross the threshold, and we can go back to being able to afford ram and hard drives again..
It used to be a handy tool - yeah, you need to vet what it's churning out, but if you know what the answer should be, it used to make things a bit quicker than writing everything out yourself.
Now though it's fallen off a cliff, makes changes to areas not requested (or needed for the request) breaking functions and reading excessive amounts of the code structure unrelated to the needed query burning through credits. Re-reading and losing context of a chat, going off investigating unrelated code bases when it's just fucked something up rather than just looking at what it did with the previous commit.
I used to be able to get a decent amount of work done, now I get five queries, four of them pointing out where it went wrong, before I'm out of credits for four hours - which is another absolute pain, four hour windows means two thirds of the credits are reserved for when I'm not at work and can't be used - at least give me a daily allowance I can burn through in the office, especially if you've adapted your tool to be more wasteful.
Very close to cancelling and just going back to doing it myself.
Does this mean the horizon stuff they push in front of the gui will take a back seat and let me use my quest 3 as I want to?
It's actually pretty good tech and great for the handful of niche's it does, for an hour or so (after that my eyes ache from focusing so close).
Fun toy, nothing more, special kind of idiot to have spent so much on something so obviously limited.
None of these will take off.
Biology is the problem. I have recently split with my partner, they kept my 100" 4k tri-laser ultra short throw projector television so, when sat in my new apartment using my third monitor (a 48" gigabyte oled, nice screen tech, good size) it felt oh so small to be starring at, I missed my tv.. so, I broke out the Quest 3 to use the faux big screen to watch my movie instead, I could sit back and enjoy the equivalence of watching a 100" screen again - even better, I could go sit in the bedroom to watch it that has AC, unlike the living room.
And, two hours in, my eyes ached - I need glasses, I have prescription lenses in the Quest, and yet, still painful to be wearing for so long.
That's the Achilles with this tech, no matter how much they try and sell it, it's a screen stuck to your face and your eyes simply aren't built to be staring this close for this long, we aren't going to be working an eight hour day in one, we aren't going to be replacing anything with tech that makes us feel unwell - all these announcements are to spruik share prices or dupe the gullible in to handing over investment money they'll never see a return on..
I've worked in systems automation for over two decades, people don't want it.
You're like, look, you can do things this way and it will save all this time and effort, freeing you up to do other more interesting things, people smile, nod, thank you, and go back to using excel.
AI agentic agents are nothing different, people are ingrained in what they do and dislike change, even if the change benefits them.
It's a fundamental mis-understanding of the workforce, and mangelment can't do anything about it, as they still rely on the meatbags to correct what the agentic will get wrong...
Easy to steal, yes, but, why would you?
I once shot a music video for a mates band in Whitby, we spent a good twelve hours there shooting in various parts of the town, having something to eat, bit of busking.
Walking back to the car parked in a huge field with hundreds of other cars I'm patting down my pockets wondering where my keys are, damn, did they fall out whilst I was rolling around on the floor trying to get a good angle? As the car in front of mine finishes reversing out and pulls off to show, nope, left them sticking out the boot!
My adhd ass was so distracted that I'd locked the boot after getting the gear out that I simply forgot to take the key out and take it with me. Twelve hours, sat there with the key in for the world to see (not even a little out of view like down the side in the drivers door like I'd done in Tesco car park one early morning), no one had taken it with the key there invitingly, don't think anyone's going to go through the hassle of actually trying to break in..
My boss is very pro-AI and it drives me a little to the job advertisements as he thinks I can suddenly do the work of ten people because he's paid for a basic claude licence.
Recently he's been making a website of he and his pals sports competitions, and as he knows no code, vibing it with chatgpt, up until, after a big competition, it wiped the database, leaving him to pay to have the database recovered by the platform supplier (as he didn't know about the auto-backups in cpanel where he could do it himself, and apparently neither did his mate chatgpt). Sadly this hasn't caused a mote of reflection on my duties..
To the article though, I'd've thought the data would be recoverable, yeah not in the recycle bin, but sounds like it was just the toc that's gone and any good scan and recover app should pick it all up.
I worked at a marketing & promotions firm (primarily - I was in a market research team) that was all Mac, it was a fight to get a windows install on mine (as I needed some windows specific software for my role) and that absolutely kneecapped the laptop - touchpad features reduced to a single click from all the fancy wankery it did, unless you paid forty bucks for a third party app that put the drivers back in, problems with the display, keyboard mapping and power management. It really didn't want to have windows on it, they really should've just let me byod, I knew more than their IT head anyway (who recommended me for his job when he left, but f'that, what do I know about managing a fleet of macs...)
It was about the only feature I quite liked in Vista, I used to have a HTPC back then and it looked quite slick having a running waterfall as the wallpaper - vista media centre actually looked quite polished too and the epg one of the nicest I've seen to this day.
I mean, I get the rest is hot garbage and most of MS stuff is, but, you know, that one little light in the dark, even if frivolous.
These days I just rock a chromecast as that's enough for plex in 4k so no need to waste the watts on a HTPC anymore, so, no use for it now, but I liked it then.
I was accidentally signed up, using the website on mobile I fat fingered the very close buttons and that was all, one click, I was in.
Took the free delivery during the free trial and then cancelled prior to being charged, so, more fool them for making it too easy to sign up, but least cancelling is relatively straight forward (once you click through all the bait of "are you sure you want to lose access to this thing we've not actually mentioned you have access to before?"...).
If you're deleting the old website to add the new, that means downtime, you have two directories "current_website" "new_website" and you update the host redirect to point to the new site, migrating with zero downtime...
Then you can do what you like with "current_website" when you've had chance to check "new_website" worked fine.
(and because I spend too much time on Reddit, before the pedantry, you wouldn't actually use "current_website" "new_website" as names...)
It's been a personal mantra for the past decade as pretty much every time I've come to a janky Excel implementation there was a more elegant solution that was easier to do, quicker and more maintainable.
Yet, just the other week, I created a spreadsheet that fetched data from two different systems (one client, one internal) and then sent SMS reminders to employees, so, you know, don't follow my own advice much...
When I upgraded from my Xiaomi 12s Ultra to the 14 Ultra (mostly to get a global rom, the phone was still fine, even at two years old) I was a little surprised to find it was actually thicker.
Not to any degree that matters, it's still a relatively thin phone (9.2mm) but the fact they went backwards I feel is a great step, showing they prefer function over this ridiculous ever shrinking form.
But the size back in to it and fill it with battery, phones could be double the current depth, makes little difference, but doubling the battery capacity, I'd buy that.
(they don't even need to double it, just push out to be in-line with the camera bump)
As a paid customer (pro plan) I got this pop-up, ensured I hadn't opted in, went to my Settings > Privacy and found I was 'accidentally' (presumably) still opted in, so I'm definitely unchecked, reloaded, still opted out, will check again in several days.
Would suggest anyone else might want to just double check in case they made an "oopsy", after all, when you vibe code you don't necessarily get what you thought you did...
I was worried this might be something more bothersome than the text I scan past - just switched to Claude from ChatGPT after last fridays update screwed Codex from being vaguely useful to down right annoying.
So far Claude's seemed more on the vaguely useful side, but I have only been using it half a day...
The changes in Codex since GPT5 came out is astounding, so much for reductions in hallucinations, every single query is now mis-understood or it changes things that were fine and didn't need changing.
Can they put a "use whatever Codex was using pre-gpt5" option?
As things stand, Intel is unlikely to have a competitive GPU or AI accelerator for at least a year, and that's assuming that Intel's GPU team survives the next wave of layoffs.
I picked up an A770 for my work rig and it was dire, unstable, glitchy, a driver update lost my third screen (support were absolutely useless even when provided all the information and pointing out where the driver change broke things) that I eventually pulled it out and got a 5070. The driver updates being bigger and bigger steps backwards in terms of stability was the last straw (bsod so hard I had to re-install windows).
Stuffed it in my NAS, figure I could at least get some value out of it for hardware transcoding of video, and nope, kept locking the system up and preventing me logging in, absolute waste of sand, sits on a shelf now as a reminder to never buy a v1 of anything.
I've been playing some games against it lately to track how well it's reasoning and deduction are going - not well.
A favourite is to play "wheel of fortune" with it, I'll spin up an online game of it and feed it a screenshot and a reminder of the rules (no adding/removing spaces, no altering of the letters already in, the clue is x etc..) and. the. amount. of. times. it. breaks. one. of. these. basic. rules. is. astounding..
These aren't something I'm expecting it to know by context of the prompt, they're spelled out right in there and it can't hack it, just apologises and promises to do better. but then immediately doesn't.
I get it's not actually intelligent and pretty much just a really big complicated search, but even by those standards, it's ignoring half the search term.
Absolutely outstanding that people are willing to trust it to do bigger and more important things - had a call this morning with the boss wanting to replace our call room people with it, good fucking luck mate...
I wouldn't fret, with such a none announcement it sounds more like a share price pull, we'll announce we're doing *something* with ai and watch the stock go up and then wait for everyone to forget, they're just not even bothering with detailing what the *something* could be, if only there was an app that could summarise a few ideas for them...
I don't imagine smaller businesses would attract ransomware fraternity
That's the thing with modern automation and crawlers, you don't have to be an attractive target, they'll just set the systems going and if they find a known vulnerability you've not patched you're popped. There's actually been a big pivot from targeted exploitation to a wider scatter gun approach, even if only a small percentage pay out, if you hit enough, it's worth it and sadly, it seems to be given the numbers seen.
