Welcome back to the ongoing saga of the Dutch government’s brilliance with AI.
After the Toeslagenaffaire (the child-care AI profiling scandal), you might think the folks in power had learned a lesson or two about the dangers of misusing AI.
They didn’t. Otherwise I would not pay any attention to it.
And this time, it is the Dutch public employment service “UWV” which has crossed the line, and has stepped into the spotlight with their own little disaster: an AI based job-matching tool that not only fails at its job but also feeds on illegally scraped data.
Because, apparently, when the Dutch government decides to screw things up, they go all in.
And to support the people trying to stop this ordeal to continue, we want to help them with a petition. We have around a 100k readers a week, and if we all chip in, this will get on the national agenda!
If you want to jump to the petition right away, click here…
Before we start!
If you like this topic and you want to support me:
- Comment on the article; Google appreciates that and it will really help spread the word 📢
- Connect with me on Linkedin 🔗
- Subscribe to TechTonic Shifts to get your daily dose of tech 💉
- Sign the petition to stop the NL Government from using our personal data to train AI ✒️
AI because it’s trendy
Let’s start with the UWV.
For those lucky enough not to know, the UWV is the Netherlands’ public employment service.
They’re the people tasked with helping job seekers find work and paying unemployment benefits. I am glad they are here for this task and generally they do their job pretty well.
And like any other organization these days, they had decided that matching people with jobs required the magic wand of AI.
But not just any AI.
This one comes with a price tag of 65 million euros.
That’s right 65 million coins.
Now, what does this high-tech marvel do, you would ask yourself…
Well, supposedly, it matches people with jobs.
But in reality, it’s more like a lottery system with a really expensive AI buzzword slapped on top. Oh, and did I mention that the original system cost a fraction of that amount? Yeah, but who cares about that when you can say you’re using AI, right?
Let’s break this down. 65 million euros for a system that doesn’t work.
The original job-matching system, by the way, cost a fraction of that. But that system didn’t have the word “AI” attached to it, so it was clearly inferior.
When you add AI to the mix, prices skyrocket, common sense goes out the window, and you end up with stealing data, like big tech these days is known for.
Let’s be real.
Did UWV pay 65 million because the system was revolutionary
Of course not.
They paid because someone said “AI”, and the tech bros in the room started nodding like it was the Second Coming.
The great data heist.
And LinkedIn was on the menu
Now, here is where things get interesting.
How does an AI actually work.
Well, it needs data to be trained, and where better to find that data than the resumes and profiles of unsuspecting people?
Yes, 8Vance, which is the company behind this AI system, thought it was a great idea to scrape LinkedIn profiles and resumes from the internet all around the world, without asking anyone for permission.
If you’re wondering, “Wait, isn’t that illegal”?
Yes, yes, yes it is.
But when you are in the business of building AI tools that cost millions, who has time to worry about pesky things like data privacy? So to get to the money quick (and dirty), the resumes of millions of people all over the world were hoovered up without so much as a heads-up, all to feed the machine.
Because what’s a little illegal data scraping when you’re revolutionizing job matching, right?
And it is not that they did not know it was an illegal act. The organization TNO (a government subsidized technology research organization), has published a document in 2021 (!) on how to train your AI without being illegal. Download it here: TNO-2021-technological.pdf
If you want to jump to the petition right away, click here…
Scraping data and desperately spinning the story
So, let’s talk about 8Vance, the supplier of this disaster.
They’re the ones who built the AI tool that UWV is now so proud of.
And they’re also the ones who initially decided to use illegally scraped data to train their shiny algorithm. But don’t worry, 8Vance is now desperately trying to mitigate the situation. They’re spinning PR faster than their AI processes resumes, hoping no one will notice.
But we notice, 8Vance.
We see you.
For those keeping track, this is the same kind of disaster we saw with the Toeslagenaffaire—you know, where the tax department used an AI system to profile families, ruined thousands of lives, even pushing some folk to commit suicide, and then acted like it wasn’t their fault?
Well, here we are again.
Another AI, another mistake, and another government agency too embarrassed to admit it.
This time, people are not getting killed, but that is not the point. Governments must stop AI for profiling, stop using illegally obtained data for training, keep humans in the loop, be accountable for the AI, and more….
Job Matching by AI
The whole point of this 65-million-euro investment is to help match people with jobs.
But here’s the thing, apparently it doesn’t work as it is supposed to. The AI tool is about as good at matching people with jobs as I am at predicting the weather. Sure, it spits out suggestions, but those suggestions are about as helpful as a fortune cookie. If you are being told that your years of experience in finance make you a great fit for a cashier position at the local supermarket.
Thanks, AI.
And yet, this is what UWV spent millions on. A tool that matches people to jobs with all the accuracy of throwing darts at a job board in the dark.
But hey, at least it’s “AI”, right?
The silent treatment by UWV and the Privacy Authority
Now, what could be worse than scraping people’s data illegally?
Not being held accountable for it.
Both the UWV and the Autoriteit Persoonsgegevens (the Dutch Privacy Authority) are sitting silently in the corner, hoping this all blows over. Remember when the Toeslagenaffaire hit, and there were promises of transparency?
Yeah, neither do they.
Apparently, nobody at UWV or the Privacy Authority learned anything from the last AI-driven scandal. If you were hoping for some accountability, I’m sorry to disappoint. They’ve gone for the “let’s pretend this isn’t happening” strategy instead.
Classic.
I am waiting for the movie to feature soon….
The damage control that’s not fooling anyone
Meanwhile, their supplier 8Vance is in full damage control mode.
They are desperately trying to convince everyone that this is just a minor hiccup. But let’s be clear, people, illegally scraping data isn’t a hiccup. Nope, this is a massive breach of trust and privacy.
They are also hoping tjat you will forget about the illegal data scraping and focus on the shiny AI tool they’ve built. Sorry, 8Vance, but we’re not going to let you off the hook that easily.
The Don Quixote of job matching
Enter Marc Drees, the lone crusader who is trying to shine a light on this mess.
Drees is like the Omtzigt of job matching (the guy, who brought the children-scandal to justice). He is a recruitment, and matching expert, AND an investigative journalist all in one.
He has been tirelessly fighting to expose the UWV’s misuse of AI and 8Vance’s shady practices His blog RecruitmentMatters (the original site = here), is of couse all about recruitment and technology, but once in a while, he tries to put some new coals under the kettle and gets the boiling going again.
He has been raising red flags, but just like in the chilren-scandal, the government is completely ignoring him. Silence is apparently their go-to move when they’re caught with their pants down.
But Drees isn’t giving up
And thank goodness for that. Because without people like him, the whistleblowers, this mess would just be quietly swept under the rug.
If you want to jump to the petition right away, click here…
A disaster waiting to happen
Why does the Dutch government insist on using AI without understanding it?
It is similar to giving a toddler a chainsaw. Nothing good is going to come of it.
From the children-scandal to this latest UWV debacle, AI in the hands of the government has been nothing short of a disaster.
They are using systems that make life-altering decisions without any oversight, accountability, or ethics. The UWV’s AI tool is just the latest example of how bad things get when you let the government play with technology it doesn’t understand.
The hero we need (but the government ignores)
This is where Human-Centric AI comes into the picture.
If we would have had HCAI in place at the Tax Department, and the UWV, this disaster could have been avoided. HCAI is all about transparency, accountability, ethics, and human oversight.
The three things that were sorely missing in this situation.
AI systems are designed to be explainable with AI. They provide reasoning that humans can understand and question. If UWV and 8Vance had followed HCAI principles, they wouldn’t be in this mess. The illegal data scraping wouldn’t have happened, because UWV would not dare to enter into an agreement with an organization like that, and the integrity of the organization would have remained intact.
But instead, we got a system built on stolen data and no oversight.
Great job, everyone.
Thanks Marc.
The Petition.
It is time to hold UWV accountable
It’s time to say enough is enough.
So to support Marc Drees in his quest, we have launched a petition to demand accountability.
Because UWV (and 8Vance) should not be allowed to scrape people’s data illegally, and AI systems should never be built on a foundation of unethical practices.
We cannot ignore the fact that “what’s done is done”, just shrug it off and say 8Vance has atoned published a few mea culpas and now back to business, right?”
Governments should not be allowed to play fast and loose with our data and livelihoods. They should not stoop to the same level of criminality like the big AI companies. They should be above that and play fair game.
Ask for our permission first.
That is something that even Apple has learnt – they are not scraping without permission, to train their AI’s. And if an organization like that can do it, surely the Dutch Government should be able to.
The children-scandal showed us what happens when AI is misused, and now UWV is proving that they’ve learned nothing.
So please sign the petition and help stop this madness before it gets worse.
If you want to jump to the petition right away, click here…
That’s the full breakdown, people….
If this isn’t a case for HCAI, I don’t know what is.
For more information, read some influential magazines:
- 🔥 Scraping is almost always illegal, says Dutch DPA
- Dutch web-scraping guidance warns businesses of GDPR breach risks (pinsentmasons.com)
In Dutch:
- UWV heeft AI-tool aangekocht die mogelijk niet aan AVG voldoet – IT Pro – Nieuws – Tweakers
- UWV has purchased an AI tool that may not comply with GDPR | AllInfo
- UWV heeft AI-systeem aangeschaft dat mogelijk privacywet schendt | Tweakers | NU.nl
Well, that’s a wrap for today. Tomorrow, I’ll have a fresh episode of TechTonic Shifts for you. If you enjoy my writing and want to support my work, feel free to buy me a coffee ♨️
Think a friend would enjoy this too? Share the newsletter and let them join the conversation. Google appreciates your likes by making my articles available to more readers.
Signing off – Marco
Ps. Join the TechTonic Shifts community and get exclusive articles.
Leave a Reply