The Rise of ChatGPT: How I Turned a Lost Case into Success Against Dutch Employment Agency.
Your chances are slim,” they said; “the legal system is too rigid, and you’re too late” — these were the responses from the legal experts I paid, but they were wrong.
When it comes to labor law, things always seem straightforward — until they’re not. What looks like a simple case can quickly spiral into something far more complex, demanding not just logical thinking, but creativity and unconventional solutions to find the desired outcome.
I have been working on an algorithm that optimizes decision-making in legal cases. The core value of this algorithm is making decisions based on information and facts while continuously updating and reevaluating these facts as new details come to light. As I worked on this, it became clear just how critical fresh insights and data can be in changing the probability of achieving the desired outcomes. That’s where ChatGPT came in. I began using it to comb through documents and refine the reasoning behind the case’s key points. In the end, this process became a stepping stone in creating a framework for what would later become Retrieval-Augmented Generation (RAG).
Determined to find the missing pieces, I started my own investigations and went through all legal articles that could shed light on the case. With ChatGPT, I transformed a bleak outlook into critical arguments that changed everything. Arguments that once felt weak and fragile were sharpened into winning points. This is how, against all expectations, I turned a hopeless case into a victory.
To give you some context:
The case is about my contract being terminated illegally, which threw my life into chaos overnight — no income, no work permit, and no insurance. Desperate to make sense of it, my first instinct was to legalize my staying in the Netherlands on the basis of protecting my private life and then apply for unemployment benefits through the UWV, the Dutch unemployment agency. But my application was shot down almost immediately. The reason? In summary, I had no work permit (not the fault of me). Without it, the individual isn’t considered “available for work,” making them ineligible for benefits. However, UWV assured me that once the work permit was obtained, benefits for those months could still be claimed. It was a bureaucratic trap, the kind where every turn seems to lead to a dead end. Eventually, I resolved the work permit issue, which should have marked a positive turn, right? But when returning to the UWV with this newfound eligibility, another excuse emerged: “Sorry, it’s too late now. You should have applied sooner.” Baffling, isn’t it? How could one apply without a permit, and yet once it was obtained, it was supposedly too late? It felt like being trapped in a Kafkaesque nightmare, where every logical step only dug a deeper hole. Seeking advice from some legal experts didn’t help either. They simply brushed off the case, repeating what UWV decided.
Here is the list of the arguments by UWV:
1. The main argument rested on a principle: “The appellant was not available to accept work within the meaning of the Unemployment Insurance Act (Werkloosheidswet, WW).”
2. No new facts and circumstances were provided on the second application.
Based on these two arguments, UWV rejected applications in March, June, August, and September 2023. UWV’s legal basis is Articles 17,17a, 16a of the Unemployment Insurance Act (Werkloosheidswet, WW), Articles 7:3, 7:11, and 4:6 of the General Administrative Act.
Well, that didn’t convince me. I knew there was more to this. The legal basis is clear, and there were supporting facts that applied to the situation. This is when I used ChatGPT to make my case.
Breaking Through the Legal Maze
On one hand, I was cautious about turning to ChatGPT for help. One reason was that I had to take the lead, pointing out the search in the right direction. What if something was missed? After all, I am not a lawyer — just someone who enjoys solving puzzles and has succeeded in uncovering financial fraud in the past. I am also not the type to take “no” for an answer without exhausting every avenue. On the other hand, the elements of the case were well-defined, and by breaking down the facts and circumstances, it felt like a path to the right solution would emerge. Hence, I started sharing my reasonings with ChatGPT, thinking through the core legal aspects of unemployment benefits and eligibility. The exploration took me deep into administrative law, labor law, civil law, and similar cases.
Here is what I found and uncovered using ChatGPT:
So, my first step was to dig into the Dutch Unemployment Act (WW) for the definitions and core principles governing unemployment benefits. These are the key criteria for receiving unemployment benefits:
“You are only entitled to unemployment benefits if you lost your job through no fault of your own. Furthermore, you must have worked at least 26 weeks within the past 36 weeks. You must also be immediately available for a new job and remain in the Netherlands.”
In my case,
- I was employed for over a year and ten months.
- The contract termination was an unlawful act, occurring a full year before the official end date [this is based on the company’s failure to get the work permit].
- And most importantly, I stayed in the Netherlands legally and was fully available to work.
What is important in this fundamental definition is that it doesn’t mention any specifics regarding work permits. Neither in criteria for unemployment benefit nor in any other employment or civil laws. Based on this, I searched further for cases with similar stories. Thanks to ChatGPT, there are plenty of cases in Rechtspraak.nl that I went through and compared with my case. Specifically, this example was also picked up by the court, later.
So, what the legal framework of these cases suggested was that:
“A person can be available for work and perform certain tasks, even illegally (without a valid work permit), simply by being present and able to work in the country (NL).”
There’s a subtle loophole here: “availability” doesn’t strictly hinge on having a valid work permit. In fact, one could still be available, even for illegal work.
Here, I was right and won the argument.
The appellant’s residence status — the termination of the WW benefit cannot be based on the lack of availability in the relevant period without further investigation.
If they are ready, willing, and able to work, they may argue for their eligibility for unemployment benefits. This could include being actively seeking employment, even if that employment would be illegal due to lack of a work permit.
Above all that, I applied for several job positions after the termination of the contract spontaneously, which shows my dedication to find another job. It also fulfills the obligation of the employee to receive the benefits, which strengthens the willingness criterion in the eligibility definition.
Further, I explored more on the legal obligations of UWV, I uncovered a significant oversight by the UWV: under Dutch civil law, employment regulations, and UWV mission (specifically Strategy 21–25), UWV had obligations they clearly neglected.
This strategy outlines several duties that the UWV must uphold, such as supporting workers, preventing unemployment, and safeguarding employees from employers’ misconduct. It wasn’t just about the eligibility and missing deadlines — they had a responsibility to provide further assistance, and they fell short.
While the content of strategy 21–25 sounds promising, 6 out of 9 issues affected me personally but were never addressed by UWV.
[A summary to this strategy provided in this link]
I also argued that I was wrongly not heard in the objection phase and UWV did not provide me with the information needed regarding that. This is also accepted by the court and UWV acknowledged that the objection was wrongly waived.
After months of back-and-forth appeals, the verdict finally arrived — and my appeal was well-founded. The court ruled that the UWV had failed in its duty by denying me assistance after the unlawful termination, and by not doing a proper investigation on my availability for work. Hence, UWV had no legal grounds to reject my unemployment benefit.
The legal grounds provided for winning this case:
- Article 3:2 and 3:46 of the General Administrative Act: these articles help ensure that administrative decisions in the Netherlands are made with diligence, transparency, and accountability.
- Article 17, 17a of the Unemployment Act: these articles ensure that individuals receiving unemployment benefits actively work toward finding new employment and follow the rules designed to encourage re-entry into the workforce.
- Article 16 of the Unemployment Act: This article provides the foundational requirements for receiving unemployment benefits under Dutch law.
I also have to mention that, due to some circumstances, I was two days late in filing my objection to the Court of Appeal. However, I did not give up; I explained my reasons for the delay, and the court accepted my arguments referring to the recent case law from the Central Appeals Tribunal.
The law is designed to protect us, so I encourage you to have faith in the process and take actions.
Conclusion:
I’ve learned that the real challenge lies not solely in intelligence or meeting deadlines, but in maintaining integrity and believing that there’s always a solution, no matter how buried it may be. Technology is here to assist us as a co-pilot in gaining more insights. With ChatGPT, I was able to dig deeper, challenge conventional wisdom, and identify the loopholes and legal foundations that ultimately turned the case in my favor.
The legal field often relies on subjective decision-making, where only well-experienced and dedicated experts can deliver fair judgments. Unfortunately, these subjective decisions can lead to flawed outcomes for the clients, especially in moments of crisis when emotions cloud judgment, making it difficult to find thoughtful solutions. In such instances, having AI-powered tools that analyze legal precedents and options as a second opinion can provide clarity and objectivity throughout the process.
Reflecting on my experience as an auditor, I uncovered fraudulent activities in a hospital’s drug stores, which led to significant repercussions for the financial manager and the CEO. It all started from a simple mismatch of invoices, but the problem was much deeper; the amounts involved were substantial and impacted not only the hospital but also several other organizations. The complexity of that situation, which required connecting dots and a keen eye for detail, relying solely on logic and experience. However, it would have greatly benefited from the efficiency that AI brings to the table, helping to refine my arguments in a more time-efficient manner.
This journey also boosts my confidence in the decision-making algorithm and Retrieval-Augmented Generation (RAG) solution that I am developing. As I continue to investigate other aspects of this case, including tax fraud, I look forward to sharing updates with you soon.
So, the next time someone tells you that something is impossible, that the window has closed, or that the rules are set in stone, remember this:
there is always a way forward. If a door is not available, you can build one. AI makes it easier to achieve that.
Feel free to contact me on LinkedIn if you’re interested in or want to learn more about algorithmic decision-making in legal tech.
For updates, follow me on Medium.com
