The Evening Standard

[Jobbo]

It had passed me by that the Evening Standard had ceased to exist as a daily print publication. It became weekly last week and is now just called 'The London Standard'. Launching today with a very AI-heavy first issue: https://pressgazette.co.uk/news/weekly- ... ront-page/



In addition to the AI generated cover image, there's an AI written article in the style of Brian Sewell. I am sure it will be riveting reading for bots. The cover asks: Can Labour really own the tech revolution? Well since that happened about a quarter of a century ago and anything since has been incremental and pretty much independent of location, it's a trite question.

Maybe this is a good place to say AI annoys me immensely and simply makes me question the veracity of anything I find online now. It's bloody annoying how all the legal service providers seem to be trying to sell it constantly.

[Beany]

For reference, the way AI gets it's data is to scrape the shit out of anything on the internet.

I run library service websites, which usually get three or four concurrent hits at a time.

AI scrapers will throw dozens of IPs from multiple disparate ASNs (so hard to block) and hit dozens of times a second, crippling the site.

Yes, I know CLoudflare exists, but enabling it impacts the staff side where you can write (sandboxed) SQL queries for running reports and the like, so it's not as easy it sounds. And these scrapers gleefully ignore robots.txt. And you can't block based on supplier because a lot of them use AWS, and other major providers that our clients also use.

The day that legislators put their heads together and throw this bunch of venal, mass-copyright-infringing, lazy grifting cunts in the bin - ideally bankrupting Nvidia, the biggest supplier of GPUs used for this, too and a bad faith industry actor for a decade now - the better.

[Jobbo]

For ~25 years the internet has been collecting the distilled knowledge of experts in forums etc and has been a great resource; if you look in the right place (which isn't usually too hard to find) you can generally find an answer to pretty much any question or solution to any problem.

AI takes that and smears it with incorrectness by amalgamating all the noise and opinion from the rest of the internet. Until AI can distinguish between the value and authority of each source, it just makes things worse and brings knowledge down to a lowest common denominator.

[KiwiDave]

I run library service websites, which usually get three or four concurrent hits at a time.

AI scrapers will throw dozens of IPs from multiple disparate ASNs (so hard to block) and hit dozens of times a second, crippling the site.

Yes, I know CLoudflare exists, but enabling it impacts the staff side where you can write (sandboxed) SQL queries for running reports and the like, so it's not as easy it sounds. And these scrapers gleefully ignore robots.txt. And you can't block based on supplier because a lot of them use AWS, and other major providers that our clients also use.

I'm not quite sure how it happened, but your day job and mine are almost the same. We've been getting hammered with them and implementing AWS WAF has been the saving grace. We can now run the sites, dealing to those who get past it, whereas before it was like trying to catch fresh air with a sieve. Conversely, our respective clients are such a good target because they're actually such a great source of reliable, actual truths seeing as they reference real physical histories.

I'm going for a sit down before I have to admit I'm Beany...

[Simon]

You need a proper bot management tool to deal with them, effectively, not a WAF.

[dinny_g]

I'm going for a sit down before I have to admit I'm Beany...

you're a far better photographer...

[Beany]

@KiwiDave You need to get way worse at taking pictures before you can be me, lad

You need a proper bot management tool to deal with them, effectively, not a WAF.

And yeah, Simon, there's some tools relating to that (CF released an AI bot management toolkit recently which looks relevant - we're testing it on some smaller systems/happy canaries) that we've been looking into but the nature of how our software works means it's not a safely fire and forget solution - we've been on and off trying to get Cloudflare WAF in front of our systems for some months now, and keep coming across problems.

I'd expect a pure bot management tool would be more useful.

I'm more annoyed that two years ago it never crossed my mind that this level of widespread, industry-scale abuse - and it is pure abuse on every important level - would be acceptable and basically ignored by any kind of regulatory body or 'big boy' tech company who might otherwise be expected to be a good industry partner and shame the actors into behaving themselves.

Again, fuck 'em all.

[Jimmy Choo]

I remember coming out of theatres in London's glittering West End and hearing the cry of "EEEEEEEEng Staaaa-rrrrrrrrrrrrr" from the paper sellers. I've never read it.

[jamcg]

So Beany, how long until every website has “accept cookies” followed by the I’m not a robot “click the motorbikes” thing just to gain access

[Jobbo]

I remember coming out of theatres in London's glittering West End and hearing the cry of "EEEEEEEEng Staaaa-rrrrrrrrrrrrr" from the paper sellers. I've never read it.

It was probably the sellers’ noises which made me buy it back in the 90s when I started working in London. Felt like much more relevant newspaper than the mainstream dailies. Loved it. I even had a bit of a short-lived thing with a short-lived tv critic from it.

[Beany]

So Beany, how long until every website has “accept cookies” followed by the I’m not a robot “click the motorbikes” thing just to gain access

That would be more of a question for @Simon, I just work here mate.

[Simon]

All of our bot management tools are transparent. You wouldn't know they were there. Noone likes even the 'are you a robot?' friction of even googles reCaptcha. We give the customer a choice to fall back to it if they want to but most don't. Besides, they are obviously easily bypassed by bot factories (cheap third world labour employed to pass a captcha whilst the bot then continues the rest of the scraping work or whatever).

Anyway, with bots you don't normally just block them like with a WAF. A bot who's blocked will realise and adapt to evade the block. You need to manage them. Eg, the most fun I had was with a customer whose competitor was price scraping them and then setting their own prices automatically with the data. We used our tool to serve fake prices when it detected the scraping. The result was the competitor trying to sell their products 10000% over retail.

[Simon]

Oh by the way @Beany, legislation is never gonna stop this, even if it were enacted. It would be like outlawing viruses or hacking.

[Gavster]

Maybe this is a good place to say AI annoys me immensely and simply makes me question the veracity of anything I find online now. It's bloody annoying how all the legal service providers seem to be trying to sell it constantly.

We’ve had v1 of this back in the blogging days. A cooking technique would get widely shared as ‘the best’ way to do something on a major blog and then everyone else would jump on it and copy precisely the same information. Suddenly Google would return pages of blogs detailing one ‘best method’ even when in professional kitchens around the world there would be loads of different techniques. It became an echo chamber.

But I get what you mean, I’ve been reading/watching stuff online and now have an inbuilt AI-radar that starts twitching when the copy is soulless.

[Beany]

Oh by the way @Beany, legislation is never gonna stop this, even if it were enacted. It would be like outlawing viruses or hacking.

Let a boy dream, Simon.

[jamcg]

As you lot are experts in this, what’s the point of the “I don’t give meta permission” bollocks the celebs are posting g on instagram? Genuine prevention, collective clutching at straws or gearing up for possible future law suits?

[Beany]

I don't use instagram, so I'd need an example

[240PP]

As you lot are experts in this, what’s the point of the “I don’t give meta permission” bollocks the celebs are posting g on instagram? Genuine prevention, collective clutching at straws or gearing up for possible future law suits?

It sounds like the same bollocks that people used to put on FB. ie: “I do not give permission for FB to use my data/photos” etc. Despite clicking ‘Yes’ to that exact thing when they signed up. It’s the contract everyone agreed to but never read.

[Gavster]

Oh by the way @Beany, legislation is never gonna stop this, even if it were enacted. It would be like outlawing viruses or hacking.

I assumed you were talking about big businesses such as OpenAI who are scraping for info. If there’s a big business doing it then surely there would be some legal recourse.

[Gavster]

As you lot are experts in this, what’s the point of the “I don’t give meta permission” bollocks the celebs are posting g on instagram? Genuine prevention, collective clutching at straws or gearing up for possible future law suits?

There’s zero point, it’s basically a signal that the person posting it is stupid.