The new age of job search – Artificial intelligence to help you find a job
News
Topics
- Income security
- Job seeking
- Membership
- Unemployment
- YTK
Looking for a job is hard work. When we hoped that unemployment funds could help their members find work, we already had in mind that job search needed to be rethought. Immediately after the law change, we were the first to try a new solution, testing the use of AI to support job search. The experiment was a success. We are more confident than ever that it is possible to combine job search and claiming income support in an effective way.
We conducted the experiment in the framework of SitraLab6 programme in cooperation with the Finnish company Headai Oy. Once again, our members played a very important role, as they participated in our experiment in large numbers.
The results of the experiment were promising. We trust and believe that AI-based targeting of job advertisements according to the skills and aspirations of applicants is feasible and will shorten unemployment cycles and improve the match between jobs and workers.
Challenges and opportunities in job search
Looking for a job can often feel complicated and daunting. We want to revolutionise this experience. Our aim is to ensure that jobseekers who apply for income support are directly matched with job advertisements that match their skills and aspirations.
This is a significant improvement over traditional job search, where you have to search for jobs from many different sources and don’t necessarily see all the vacancies. Our solution therefore saves time and reduces unnecessary and duplicated work in the job search.
Results of the experiment and learning experiences
Over 1300 volunteers participated in our experiment during the summer of 2024. Although the solution was still in the pilot phase, the results were encouraging: 20% of participants said they would apply for a targeted job, and 10% said they would find the targeted jobs excellent.
The quality of applicants’ skills data proved to be particularly important. In order for AI to correctly match jobs, applicants need to describe their skills and aspirations as accurately as possible. We already had time to test different ways to improve the quality of this data.
Not everyone found a suitable job. The most typical reason was that there were no jobs in the job search area that matched the skills and wishes. In the pilot, we also sent less suitable results to all respondents, as we did not want to limit the targeting yet in the pilot. The main reason for this was that we also wanted feedback on the targeting, which was not accurate. This information will help us to further develop the solution.
Artificial intelligence
We used a technological solution developed by the Finnish company Headai Oy. The solution is based on Headai’s proprietary language model, which treats natural language as a mathematical equation using neurocomputation. The algorithm identifies keywords related to knowledge. The application and the job advertisement are compared and a numerical value is calculated for matching.
Headai’s AI solution had a few features that supported our use case. A very important feature is that the model’s training does not use the data fed to it. In other words, the AI used does not learn from the data it analyses. This is essential for data protection. In addition, any biases in the data do not distort the model itself.
Another important feature is that the model used was not generative. In other words, the model does not create new content. This ensures that the results are always reproducible and can be better controlled.
The direction of the future
Our aim is to make the service available to all its members. Based on the pilot, a production version is now on the drawing board, which will better serve jobseekers. At the same time, it offers the possibility to speed up the process of finding a job, to the benefit of jobseekers, employers and society. Our aim is to have the production version up and running within the next year.
The future of job search looks bright and we intend to be at the forefront of this development. We believe that AI will make the job-to-jobseeker encounter more efficient, more human and less stressful.