> The Valley doesn’t understand how fed up normal people are with AI.
I occasionally hang out with normal (non-tech) people. They're either agnostic to it or they love it. None hates it, unless they have some reason to think they're going to lose their job.
HN is that one place where I routinely hear negative sentiments about LLMs.
Here it seems to depend on the topic - discussions surrounding how to get the most tokens/$ are positive with most offering ideas on how to improve their workflows. Topics that invite discussion about big labs' hypocrisy regarding IP theft/copyright infringement and attempts at regulatory capture or rent seeking behavior are more negative.
These topics attract different people, but same people can hold the two views that aren't necessarily in conflict - I am forced to use LLMs at work, and within the industry there is no alternative. I also derive a lot of benefit using it for personal projects. I would also be very wary of monetizing any of these personal projects due to being completely blind as to how/where the code used to train the models came from.. and I am surprised how slow the legal system has been to address this issue, it's the wild west at the moment.
I think groups have become self-sorting on this topic. Some furry groups I'm in are fine with it, some loathe it.
I tried suggesting to one group who hated it that they could go to a nearby "PauseAI" protest, the reaction was mostly "Bah, they only want to pause it, not to end it!"
Completely opposite for me. Have mostly non-tech family across a variety of industries. Finance, accounting, education, pharmaceuticals.
Everyone got the "use this or we will kill you" speech from management, while finding its actual usefulness (they aren't coders and weren't using it for code!) to be questionable vs the hype.
Further a lot of people got the "you better AI-max or-else" ultimatums in Q1 only to get the "you idiots why are you using so much AI this costs money" by Q2/Q3 whiplash.
Reminds me of Epic (healthcare company). I was close to finishing grad school, and they wanted my SAT scores. When I gave it to them, they couldn't stop gushing. I wanted to wash my hands with a lot of soap at that point.
In the US, businesses ask for it because the SAT serves (or used to serve) as a decent proxy for IQ. A business cannot ask an applicant to take an intelligence aptitude test without a lot of extra cost to ensure the business complies with Griggs v. Duke Power Co (1971), so the cheaper workaround is to ask for SAT scores in conjunction with other proxies such as the school into which you were able to gain admission.
Businesses can and do ask applicants to take general cognitive tests, without excess compliance costs. IQ testing is presumptively legitimate for essentially all white collar jobs. The biggest companies that administer these tests for hiring companies have logo crawls that include a few of the best known companies in the world.
The reason more companies don't use IQ tests is that they don't work well for candidate qualification. Also, candidates hate them.
> In the US, businesses ask for it because the SAT serves (or used to serve) as a decent proxy for IQ. A business cannot ask an applicant to take an intelligence aptitude test without a lot of extra cost
Epic actually did (does?) administer a full blown "entrance exam" - generic problem solving skills, with some basic coding and math stuff. They literally arrange with the university to administer it in a classroom setting. So yeah, kind of like an aptitude test.
I actually don't mind that, as it ensures all applicants are being judged on a level field.
It's quite rare for people graduating with a BS to be quizzed about SAT scores - the focus tends to be more on GPA, etc. But asking for SAT scores when you've been in grad school for so long...? Kind of ridiculous. And if you are going to ask, might as well ask for the GRE scores instead.
They also did (do?) a full personality quiz. You know, the type where they give you a scenario, and say are you more of A or B, and you often find yourself to be neither or both...
> Most of my querying can happen thru Gemini or Claude even if it's a bit slower but I don't have to pay $10/mo for something that is not worth $10 to me now.
This is more work up front, but I have a bot I communicate with via Telegram, and I've hooked up Kagi API Search and Extract, and now achieve something similar. I just ask it to do a web search to try to find the answer to my question.
I think at some point I'm going to get banned from HN for repeatedly saying this, but here we go:
This thread highlights why I often say that at some point in the future, the obsession with whether something was written by AI or not will be classified as a mental disorder.
It is tiring to virtually every day see this discussion play out in some HN thread or other. It's also just noise. I don't think your comment, nor the thread it inspired, added anything of value to the discussion.
KMyMoney stopped importing (some) OFX files for me. I asked an LLM to debug.
It was a wild journey. An obscure bug in some QT library that was triggered because of something completely unrelated that I had installed briefly in 2019 (and uninstalled years ago).
I had it fix the issue by cleaning the leftovers from 2019, but it's still a QT bug. I had it do a writeup, but I'll never submit it. It's fairly sophisticated and I don't know QT code at all.
If it can put together a minimal reproduction of the bug I hope you might submit that. Tossing a commit full of slop over the fence is a dick move, sure, but tossing a small script over the fence that has approximately no dependencies and reliably reproduces the bug is pretty much guaranteed to be helpful. Leave it to the experts to figure out how to actually solve the problem.
I never fixed the bug. I just removed the thing that triggered it.
I believe I do have some sort of a reproducer, but I likely wouldn't understand even the reproducer script. I think it won't be taken well if someone files a bug, says "Here's a script that reproduces it", but himself doesn't understand what the script is doing.
That logic is true of "contributions" - here's some undocumented code, it seems to fix my own poorly specified problem, I have no idea how it works, please merge.
For bug reports (at least IMO) a minimal example to reliably reproduce the error without any fuss is the basic building block. The reporter doesn't need to understand why it works let alone offer an explanation. In fact for complex projects more often than not any purported explanation will be subtly incorrect and actively misleading. If anything further "documentation" of bugs is unwelcome. Unambiguously demonstrate the defect and GTFO.
Well sure, if you use a library in a borderline incomprehensible and clearly unintended manner and encounter an edge case then perhaps it isn't useful to report it. But while obscure, the usage you described (ie encountering stale config or cache data from a decommissioned install) can hardly be said to be unintended.
If an end user encounters an error while staying within the bounds of "expected usage" then the bug is guaranteed to be valid. You just don't necessarily know whose bug it actually is (for example a broken library versus broken usage of the library by the consuming program).
> The web is getting harder and harder to use due to all the anti-bot mechanisms site operators had to put up.
Indeed. I recently set up a bot on my local machine that I can interact with via Telegram. I gave it the ability to retrieve the contents of a URL via Playwright (with modifications). Basic things like getting the Readability view and getting the site as Markdown with links.
I was essentially mimicking the feature OpenAI etc offer where it can do a web search for you and get you the answer you seek. I just didn't want to pay a company for that feature.
It's not autonomous. It only acts when I tell it to. The actual search is via the Kagi API, but the URL contents is via Playwright. I also ensured it can't do lots of requests at a time. There's a mandatory (random) sleep of between 1 and 5 seconds between two requests.
In less than a week, I've been flagged. When I do a Google search (as a human), I have to prove I'm human. Many sites have simply blocked me - I can't buy stuff from various vendors[1], I cannot access several news sites, etc.
The thing is, I didn't really access that many sites with the bot! At most it's the equivalent of someone mildly addicted to the web. Yet they caught me quickly.
Obviously, I disabled the bot's capability to get site content. I wonder how long my IP will be in the blacklist.
I know some people will say "Good riddance" to the outcome and to what I did. I'd love to have a real, serious discussion on whether what I did was immoral. But whatever your thoughts on the matter, do ponder over the fact that the system has made it hard for average Joes like you and me to have this "feature", but if you pay a giant tech company a monthly fee, you can have it.
[1] Thankfully, Amazon didn't block me. In the past I'd avoid them, but now I'm in the dilemma where I don't want to buy stuff from sites that make it difficult for me to buy stuff from them!
I think that major gateways like Google and Cloudflare and Akamai are doing things like TLS-fingerprinting and smearing.
I normally use a VPN to access the Internet. It's an ordinary consumer-grade VPN, and I use it for pretty innocuous things like watching the international versions of Netflix or avoiding being geographically-profiled and charged more when I book travel. The VPN itself, I'm sure, is being used for all sorts of naughty stuff: when I look it up in whatismyip, it comes up as VPN, "abuse detected", and "residential proxy detected". I expect to get blocked occasionally on it and just pause the VPN then.
What I found surprising is that pausing the VPN doesn't always clear the CAPTCHAs. The next request will also fail, sometimes. For a while I was wondering if I have a residential proxy on my network, but I went hunting with lsof and bandwidth monitoring and a list of all the open connections through my router, and it seems clear. Yet even with VPN disabled, I'll get CAPTCHAs from Google and Cloudflare.
I suspect they're using browser/TLS fingerprinting and then smearing the results of that across bad IPs. If you connect to them through a known-bad IP, they not only will CAPTCHA you, but they will record a fingerprint of any data they can. Then if you connect again with that fingerprint, regardless of the IP, they block that too.
Ironically, if you are actually malicious, you can get around this pretty easily. Because you'll know not to use a fingerprintable browser, or to rotate fingerprints. So you have one completely separate identity on a given IP (that is not yours), and if it gets blocked, you roll up another completely separate identity on a different IP. But this is too much work for an ordinary consumer who just wants to keep their IP private. So you end up with this excluded middle where the only rational strategies are to either do exactly what the website expects you to do (with no funny business, even if it's innocuous), or to go whole-hog as the aggressor where you're determined to skirt their policies no matter what, and if you're doing that you might as well amortize the effort by extracting everything of value you can from them.
Yeah, I didn't want to put too many details as it would obscure my point. I was actually using a modified Playwright that's been designed to better act as a human. It works, in that I got fewer captchas etc than regular Playwright, but apparently not well enough.
That modified Playwright was based on Firefox, and that's what I use for my regular browsing. So that didn't help.
I can actually launch Chrome and not deal with this - but I'm not going to. In fact, what surprises me is that I can create a new Firefox profile and not get those captchas. I don't know how my using that modified Playwright as a bot got them hooked on to my Firefox profile. I suspect they had my FF profile as a "suspicious" entity even before I made the bot (I use a number of privacy/anti-cookie extensions), and the bot pushed the score over the edge.
I'll probably start fiddling with my profile - disabling extensions, until things seem to work. Thankfully, no one else in the household is impacted.
For now, I got around all this by just paying for the Kagi extract API. It's not as good as my modified Playwright had been, though. I thought of a VPN, but I suspect that won't last long (not a residential proxy, etc), and I don't want an ongoing monthly fee. I like Kagi's pay as you go option.
But as I said, it's a weird world where suddenly Amazon is coming out the "better" site/vendor. It's funny - in the old days I was pretty anti-Walmart, but a few years ago I started preferring them over Amazon (and even several local stores) for a number of reasons (better collection than local stores, not having to deal with fakes for online orders, etc).
I've not been liking the state of the web for years prior to the advent of LLMs, but this is feeling like a final straw where I accept "the web is dead". We need a fundamentally different protocol - the model of "Set up a server anyone can reach, put content out there, and then put up a lot of sophisticated technology so your server isn't hammered" is not tenable. The protocol itself needs to manage the load.
My residential network has been on Google's shit list for a few years now, I don't think it will ever go away. Like others on this thread if I see a Google recaptcha that site won't get my business as I have to exit since the captcha is an infinite loop.
Expect to remain blocked, that's how defense works, some services plenty use will have is bot or known proxy exit node set to true, and some websites use the same defense services, request velocity is just one factor, if you don't have random human like scrolling and jitter with a bot, it's a dead profile walking, metaphorically speaking
I just did this as well. I also added paywall bypass by default along with inline article chat. The internet is such trash these days.
I made my modifications at the network level and auto redirect a configurable set of sites to go to reader mode/paywall bypass by default with a MITM proxy and a custom CA cert I trust
That is entirely ISP dependent and many don't actually work like that. Your cable modem gets a specific IP for as long as Comcast or whoever sees fit, restarting does nothing on the WAN side.
On cable ISPs, oftentimes, changing the router's WAN interface's MAC address and then rebooting the modem will get you a fresh IP address, but rebooting the modem without first changing the router's WAN MAC address will get you the same IP address.
It depends on the DHCP part and your provider lease time. When you access the Internet through a modem and landline, it's an highly dynamical system, where it's difficult to guarantee every caller/user gets its own IP. Especially when IP sharing is used among the clients. Even more difficult it is with the mobile Internet. It's a constant connection loosing/reconnecting. Actually resetting and freeing up IPs for new devices, the whole time. With cable Internet (the TV cable) one is having a kind of stationary access point. The network provider can see it's topology and it never changes once a leafe has joined the network. I typically have the same access creds over months. Even when I restart my cable modem, it's still the same IP. When I was on modem, I could get a new IP every second of I wanted. Now I have to think twice, because I just can't get a new IP, even if is forcedly, not.
Doesn't work like that, only on mobile phones you can toggle airplane mode to get a new ip assigned and cgnat makes sure mobile ips are "trusted" , but they fingerprint you beyond ip address
There are plenty of reasons to criticize Roku, but I'm still a very happy customer.
reply