Rick Mason

@rick@umbracocommunity.social
28 Followers
20 Following
29 Posts

.NET C# web developer for The Pensions Regulator and Stoolball England. Into walking, stoolball, and jive dancing.

#tootfinder

The concern here is that old search engines can give you links and you can go into them to verify the content. AI taking over our search engines hides the sources of information, forcing us to blindly trust the AI and wherever it got its information from.

#AI

We just published v6 of our #govuk Design System implementation for #Umbraco. Highlights include Umb 13, GOV.UK 5.2, accordion & task list components, 3 & 4 column layouts, file upload validators, debug symbols, and easier extension of rich text properties https://github.com/thepensionsregulator/govuk-frontend-aspnetcore-extensions/releases/tag/v6.0.0
Release v6.0.0 · thepensionsregulator/govuk-frontend-aspnetcore-extensions

New versions This release updates our officially supported versions to: GOV.UK Frontend v5.2 govuk-frontend-aspnetcore v2.2.0 Umbraco 13.x (with 13.3.2 as our minimum, because it's the minimum wit...

GitHub
Bam! Make any login or authentication journey more accessible with the new Password Input component from the GOV.‌UK Design System.
It’s even WCAG 2.2 friendly! #a11y https://design-system.service.gov.uk/components/password-input/
Password input

Help users accessibly enter passwords

Office chairs don't have to be boring.
The only solution I've found to updating the back office preview cache is to rename the HTML file, so the cache seems to be based on the name #umbraco
Anyone know the trick to making #umbraco update a custom back office preview for a property editor? Just upgrading v10 to v13 and I've got a block preview for a rich text editor which is stubbornly using a cache from somewhere...
Our GOV.UK Design System implementation for #Umbraco now uses govuk-frontend 4.8. Our UmbracoTestContext has more mocks available too. https://github.com/thepensionsregulator/govuk-frontend-aspnetcore-extensions #govuk #govukdesignsystem
GitHub - thepensionsregulator/govuk-frontend-aspnetcore-extensions: Adds client-side validation, additional components and editing in Umbraco to https://github.com/gunndabad/govuk-frontend-aspnetcore

Adds client-side validation, additional components and editing in Umbraco to https://github.com/gunndabad/govuk-frontend-aspnetcore - thepensionsregulator/govuk-frontend-aspnetcore-extensions

GitHub

Heya! 🥳 Just released TurboXml https://github.com/xoofx/TurboXml a .NET library that provides a SAX like XML parser that can parse small to large XML documents with zero allocations (apart for an internal buffer that is pooled) and is able to accelerate its parsing with SIMD! 🚀

I can now continue my prototype of rendering some SVG by using this library 😅

#dotnet #csharp

GitHub - xoofx/TurboXml: A .NET library to parse XML at lightspeed with zero allocation.

A .NET library to parse XML at lightspeed with zero allocation. - xoofx/TurboXml

GitHub

In case you missed it, the GOV.‌UK Design System published a new major release of GOV.‌UK Frontend! 🎉 It's the frontend framework that gives thousands of UK public services its look and feel.

Version 5.0.0
– introduces a support model that follows the fast-moving browser market
– removes legacy frameworks and old toolkits
– improves performance, and
– makes it easier for services to make future changes that will meet the latest accessibility rules (WCAG 2.2). 1/n

Our latest #umbraco #govuk release enables conditional validation of dependent fields. Put [Required] on this email address and it'll only fire if the radio button is selected. https://github.com/thepensionsregulator/govuk-frontend-aspnetcore-extensions
GitHub - thepensionsregulator/govuk-frontend-aspnetcore-extensions: Adds client-side validation, additional components and editing in Umbraco to https://github.com/gunndabad/govuk-frontend-aspnetcore

Adds client-side validation, additional components and editing in Umbraco to https://github.com/gunndabad/govuk-frontend-aspnetcore - thepensionsregulator/govuk-frontend-aspnetcore-extensions

GitHub
×

The concern here is that old search engines can give you links and you can go into them to verify the content. AI taking over our search engines hides the sources of information, forcing us to blindly trust the AI and wherever it got its information from.

#AI

@ned

We computer people have been warning about lack of transparency in devices, software and services for as long as I can remember.

@angelastella @ned It's not just about AI. The trend here is to put QR codes everywhere now. To subscribe to any event. To get the menu at a restaurant and most importantly to pay the bill without the waiter coming over.
How easy it is to put a sticker replacing that QR code?

@s1m0n4 @angelastella @ned

It's happened with EV charging poles. And then people not looking and just going ahead and doing payments on that fraudulent site.

@jan @s1m0n4 @angelastella @ned Saw your post too late and posted something simular, apologies.

@jan
What I don't get with QR code scams on EV chargers. These things usually have displays, don't they? Why there is a sticker?

@s1m0n4 @angelastella @ned

@flxtr @jan @s1m0n4 @angelastella @ned the amp chargers at work just have a sticker that the app looks at, so it at least has to be on their network.
@zoetek @flxtr @jan @s1m0n4 @angelastella As an ad designer I always struggled with how management wanted to hide URLs in QR codes. My practice was to present the URL with the option to scan a code if that made things quicker for you. ie. "Visit this website or scan the code below." Management always stepped in and made me strip the URL. I was a Blackberry user so I knew QR codes a decade before any of these Apple/Android types had ever heard of it, so I didn't just assume that everyone used it.

@zoetek @flxtr @jan @s1m0n4 @angelastella

This practice also leads to rampant use of junk domains and tracking strings. In my day you registered proper domains and created simple, readable URLs. Now they hide URLs under QR codes and append long strings to every link. This helps to confuse URLs and further obfuscate users from being able to look at a link/URL and quickly ascertain its legitimacy. They have created a scammer/phisher's paradise, at the behest of upper-management and corporations.

@ned @zoetek @flxtr @jan @angelastella another concerning aspect imho is that some non-techy people aren't comfortable with links anymore.
Clicking on a link is not something trivial anymore. They feel more reassured with a QR code, which is complete nonsense to me.

@s1m0n4 @ned @zoetek @flxtr @jan

We must educate them again about how human-readable identifiers add to security. Oh well.

@s1m0n4
I think typing in an URL is challenging for non-techy people these days. Maybe because modern mobile browsers hide parts of the URL (scheme, randomly some third level domains). When you tap in the address field in mobile chrome, to edit the URL, the field gets cleared…

@ned @zoetek @jan @angelastella

@flxtr @s1m0n4 @ned @zoetek @jan

Yes, browsers actually fighting us should be the first thing to fix.

@angelastella
Google'll never fix it. AFAIR they announced about a decade ago or so, they wanted completely get rid of the address bar; that URLs should be completely hidden to users. Of course they got a lot of flak for it, so they are pushing it very slowly. And other browsers do the same because their UI designers think that's how it's done now. The fall from grace was probably to combine the address bar and search bar, even though this is incredibly practical…

@s1m0n4 @ned @zoetek @jan

@flxtr @s1m0n4 @ned @zoetek @jan

I'm not counting on Google for anything. But as long as Chrome is in use and we can't replace it at least it should be properly configured.

@angelastella
Non-techy people will use whatever comes with their device. I don't really see how you can influence that. There are now people who have grown up with the Chrome browser. They don't question it. Such people are as likely to switch browsers as I am to come up with the idea of installing Danish (outward-opening) windows in my house.

(I hope I'm just unimaginative and disillusioned and that someone manages to make tech literacy hip.)

@s1m0n4 @ned @zoetek @jan

@flxtr @s1m0n4 @ned @zoetek @jan

I'm no good at propaganda either and mediocre as a teacher but I've seen changes in my life and not all for the worst.

@zoetek
Yeah, but some rando could just put a sticker with their payment link anyways. Then you might not be tricked, as you find it suspicious, that it's not working with the app. But another person might scan it with their regular QR-code app and still get scamed.

I think QR-code is ok, as long it's shown by the Monitor and the current status and next step is clearly shown.

@jan @s1m0n4 @angelastella @ned

@s1m0n4 @angelastella @ned Happened in fraudulous cases in Belgium too, at gas stations. Luckily not a wide spread event….for now.

@ned

Perplexity.ai shows you sources.
Multiple sources, images, documents.

@n_dimension @ned Still not enough as no one clicks on sources. It's easy to show spun content and put a subscription fee to it. Substantially more difficult to make that original content all AI bros are salivating over.

@sambondor @ned

I click on sources.
Not all the time, but when it matters.

Hence, your argument is flawed.

@n_dimension @ned Take a website like Reddit which is one of the largest forums on the web (artificially pushed up by Google). They even have a pinned comment which summarizes the article otherwise users will just argue over the title. That's a behavior commonly found on all social media. So, it's great that you click on some links, because most people don't unless something seems off to them, prompting the need to check the source for validity. Overall, this leads to less traffic to sources.
@n_dimension @sambondor @ned and if that source is behind a pay wall, do you sign up just to check it?

@UkeleleEric @sambondor @ned

I click on another source that isn't.

Paywalls are not an artifact of AI
But that of capitalism.

@n_dimension @ned this. It is a UI thing.

Search engines used to use link indexes, then algorithms, now AI. It is a technology incorporated in the discovery process for the search engine.

The relevant question is how the information on the SERP is presented to the user. Google will try to dissuade people from clicking on any link, because then they leave their site. A good search engine will encourage it.

@ned
This should scare everyone

@ned but there's a tiny "AI summaries can be incorrect" disclaimer in light grey text in a locked cabinet in the basement with a sign saying "beware of the leopard", so it's definitely the user's fault if they get scammed.

@hacks4pancakes

@womble @ned @hacks4pancakes
Whereas it should be a large bold message saying "LLM-Generated Responses Are Not Search Results", above and below the summary, or tattooed on the back of the user's hand as needed.

@petealexharris the tattoo should be on the hands, forehead, and inside of the eyelids of every product manager even vaguely associated with any search product in existence.

@ned @hacks4pancakes

@womble @ned @hacks4pancakes

That seems cruel. Maybe just a sign in all the meeting rooms that the engineers can tap sternly and repeatedly.

@petealexharris Nope. Tattooed the old fashioned way, with a pointy stick and a big rock. @womble @ned @hacks4pancakes
@petealexharris desperate times call for desperate measures.
@petealexharris @womble @ned @hacks4pancakes copying that quote and distributing it everywhere if you don’t mind 😇
@ned Is that Automatic Ignorance or Artificial Intelligence?

@pinhman @ned

LOL, I like that.

From now on, I am going to use

"Automatic Ignorance" instead of "Artificial Intelligence", too.

@ned @pinhman It's automated learning and it's very bad at learning things

@ned Source is everything! Search engines shouldn’t process the information the way they seem to be doing. Providing links to the source means providing means to evaluate the credibility of the data. Hiding the source (behind #ai or in any other way) means also the data is stolen and the source becomes merely raw data provider.

This is very short sighted to the search engine business as well - how long does it take until websites start to block this kind of data harvesters?

@ned the number google gives cannot be trusted for a long time, scammers have been gaming it for ages and google themselves also mitm calls for a while if I recall correctly.

@ned

really, privacy is a huge concern.

today i was checking out the google photos for as local restaurant, and some old lady had literally uploaded a photo of a note that contained the full name, phone number, address, and DATE OF BIRTH for several 'friends' and family members. like JFC. it seems to have been up for months

seriously.
i was stunned. yes, i reported the photo

but things are pretty bad.

@ned oh yeah, this has been a problem for a few years now, even before the current uptick in companies trying to shove AI in everything, and I've definitely noticed that it was getting worse

I used to work in a call center and I'd constantly get calls from people who had either previously spoken with someone who was clearly a scammer after googling the number for my employer at the time, or who got our number while trying to get the number for another company in the same industry

@ned “forcing” is a strong word. It’s not like Google doesn’t also show you search results (below the AI. And the sponsored links, many of which are also malicious).

You just need to scroll. A lot.

@mathaetaes @ned

Seeing how many people blindly accept what comes out of <insert name of LLM here> as the truth... it's already the case that it is being force-fed.

With SearchGPT that will only get worse.

@mathaetaes @ned
It is forced by being the first. Forced into labelinv is the best.
Now, Google tells you this is the best, would you scroll further? in a harry? on the phone? would you scroll further?
@ned I refuse to talk to the ai fuckers that answer most Corp help lines. I just swear at them and tell them to fuck off and give me a human… until they do. 😜
@Pineywoozle @ned
I was never able to get to human section.
@mikalai Really? I always get there. I just don’t give them any info other than that I think AI is garbage and I want a human. @ned
@ned
I sometimes use Perplexity which gives links to its sources. But I would never use a public "#AI" tool for business or sensitive stuff*. That's as naive as clicking a link in an unsolicited email.
(I do recognise that we are ALL vulnerable to some degree & according to the situation.)
(*I do use a range of paid, professional, specialised AI tools for some specific purposes that they are designed for, depending on their certifications & specific security protocols.)
@Quantillion @ned Yes, I never used QR codes or scan to pay either and keep my cards in a safe purse where they can't be read. I just take it as read that any "convenience" will have built-in fails that can be exploited.
@ned and it is known to make things up.

@ned

Das ist erst der Anfang. Wir werden noch mehr Betrugsfälle durch und mit AI (Automatic Ignorance ;-) sehen.

@ned @perplexity_ai does not hide sources. Don't use crap. Use best.
@Trabuster @ned @perplexity_ai whether it hides or shows the sources, I'm not using any kind of AI because it still uses like 10 times the energy that goes into old-fashioned searches.
@titia @Trabuster @ned Also, people aren't trying to use "the best AI confabulation machine", they're trying to use a search engine.
They don't want to have to research if the top result is real or just a string of words statistically likely to sound like facts, they just want to find the contact page of the company.
@ned ai can be trained by scammers easyly, if u prepare scam info thousands of pages not public but in the place for a ai search, ai may give weighted info regarding this hack.. ai is not capable of making the decesion of the source if it is secure or real .... there will be new job that prepares scam ai search pages to imply an ads or political support...
@ned that's why AI search always has to offer sources it's used. I.e. @kagihq does it well 👌🏻
@ChristianKrebel @ned @kagihq
Well, go to Sabine on youtube, about physics.
You look for couple of videos where she she reads articles, and then actually reads references, and often refered material isn't saying what is claimed.
refs must be read before coming to conclusion.
@mikalai you're right, refs have to be read and not blindly trusted. That's why some social pages warn you before sharing a post if you've not even clicked on the link in it. People are just too lazy