TZUTC: -0500   
   MSGID: 1507.consprcy@1:2320/105 2d2ffc7e   
   PID: Synchronet 3.21a-Linux master/123f2d28a Jul 12 2025 GCC 12.2.0   
   TID: SBBSecho 3.28-Linux master/123f2d28a Jul 12 2025 GCC 12.2.0   
   BBSID: CAPCITY2   
   CHRS: ASCII 1   
   FORMAT: flowed   
   Researchers warn that skill erosion caused by AI could have a devastating and   
   lasting impact on businesses - but it may already be too late   
      
   Date:   
   Tue, 16 Sep 2025 17:57:00 +0000   
      
   Description:   
   Esko Penttinen, Associate Professor, and Joona Ruissalo, Post-doctoral   
   researcher at Aalto University, tell us about the risks of skill erosion, and   
   why AI makes the problem more urgent.   
      
   FULL STORY   
   ======================================================================   
      
   The AI boom is changing workplaces in myriad ways that extend well beyond   
   efficiency gains. As companies automate more knowledge work, researchers warn   
   of a worrying threat: the erosion of human skills.    
      
   This "de-skilling", once seen as the natural shedding of obsolete tasks, can   
   leave employees unable to perform essential functions when automation fails.    
      
   Few studies capture this as clearly as The Vicious Circles of Skill Erosion ,   
   published in 2023. That paper examined an accounting firm where reliance on   
   automation fostered complacency, and eroded staff awareness, competence, and   
   the ability to assess outputs.    
      
   When the system was removed, the firm realized its employees could no longer   
   perform core accounting tasks.    
      
   The paper's findings are more pertinent than ever in an era where AI tools    
   are becoming ubiquitous.    
      
   I spoke with two of the papers original authors, Esko Penttinen (Associate   
   Professor at Aalto University), and Joona Ruissalo (Post-doctoral researcher   
   also at Aalto University), about the risks of skill erosion, why the issue is   
   more urgent in the age of AI, and what businesses can do to prevent it. Your   
   research talks about AI's erosion of workspace skills.   
      
   What motivated you to explore that subject?   
      
    Esko Penttinen (EP): In our research team, we pursue something that we call   
   phenomenon based or problem-based approach to research. By that, we mean that   
   we always start our research project with a practical problem that we   
   encounter in real-life.    
      
   In this case, it was a serendipitous interaction with an informant in an   
   accounting company who, in a side sentence, told us that an automation system   
   had been removed from their IT architecture, revealing that their accountants   
   skills related to the underlying business process had been eroded.    
      
   We found this observation made by the accountant fascinating and embarked on    
   a case study to understand how this skill erosion manifested in the   
   organization and how the erosion had happened. This truly was a revelatory   
   case in the sense that it is extremely difficult to get access to this kind    
   of failure case.    
      
   Organizations are typically reluctant to share their experiences regarding   
   failures. We were very lucky and remain grateful that the organization let us   
   study this phenomenon.   
      
   What does it matter for every business and beyond?    
      
    EP: Our main finding points towards the delicate balancing act of handing   
   tasks to technology while being mindfully engaged in the business process.   
   This is what we depict in the figure in our paper (Figure 3 on page 1391).    
      
   We claim that most companies need to take their stance on this mindful   
   conduction vs. automation reliance conundrum. These loops are by no means   
   mutually exclusive, but we claim that it is very easy to go for the extreme;   
   either fully automate something or then fully manually conduct the process.    
      
   The sweet spot is located somewhere in between these extremes. But this is   
   easier said than done, as is shown in our case. (Image credit: Association    
   for Information Systems) Some might say that this is just evolution and that   
   things have to change.   
      
   How is this different?    
      
    EP: This is a good question and we struggle with this. For decades (and   
   centuries for that matter), the objective of technological development has   
   been to free human effort for more productive work.    
      
   There is this pro-automation argument, claiming that we should automate   
   everything that can be automated so that human effort can be targeted to   
   higher thoughts.    
      
   This applies in many arenas, accountants should not be manually entering   
   invoice data into systems, but rather analyze the invoice data to provide   
   insights to managers.    
      
   The flipside of this is that higher thoughts cannot be had without engaging    
   in details.    
      
   By engaging in tedious manual work, employees often immerse themselves in   
   nitty gritty details about the business process. And by this immersion,    
   better insights can be gained.   
      
   Given the quasi-evangelistic nature of the current AI ecosystem, can anything   
   be done to mitigate this erosion?   
      
    EP: There are measures that organizations can take, for sure. We are writing   
   a paper on this topic, hopefully getting it out next year.    
      
   In that paper, we stress the importance of technical and organizational   
   control points. Instances of checks enforced to employees to check that they   
   are in the loop, in other words, that they understand the actions taken by    
   the automated agents or AI-tools.    
      
    Joona Ruissalo (JR): Other options are to organize recurrent workshops where   
   employees work together to solve complex or unusual cases or even build   
   automation-free training environments to raise awareness of potential gaps in   
   skills and domain knowledge.    
      
   Also, to not fall asleep behind the wheel, organizations can run periodical   
   audits in which they are asked for justification of outputs or implementing a   
   nudging feature that regularly asks the human employee to validate an output   
   with justifications.    
      
   In addition, adding explanation features to the essential systems they are   
   using in their daily work allows to not just learn or quickly recap how the   
   system operates to produce a specific output.    
      
   These measures should ideally be implemented in tandem to challenge employees   
   to engage in reflection and critically evaluating AI outputs.    
      
   Actively maintaining organizations skill and knowledge capital puts them in a   
   position where they can quickly adjust to external shocks where core systems   
   can be taken down for an unspecified time and changing technological   
   conditions to better co-evolve with new technological capabilities. One of    
   the themes we discussed via email was prompt engineering (or composition as I   
   put it).   
      
   Can you draw a link between prompt engineering and the issue of skill erosion?   
      
    JR: Getting the prompt right is one thing, but it is quite another to   
   evaluate those outputs.    
      
   These require different skill sets, but both necessitate competence in the   
   problem domain, such as in accounting or software engineering, and lengthy   
   exposure to the contextual intricacies to become efficient in composing the   
   prompt and then validating the output you receive.    
      
   Of course, you can take ready-made prompts and let them spit out a response   
   without scrutinizing them in depth, but where is the critical evaluation of   
   the outputs and active reflection on why and how you are going about the   
   process?    
      
   This is where the dynamics of skill erosion come to play: the issue of    
   relying on the pre-validated prompts made by you in the past or someone else   
   and repeatedly relying on those ceases your active engagement with the task   
   where you no longer apply your skills and knowledge to the full.    
      
   As the prompts continue to produce the desired outputs, such as accurate   
   financial information or lines of code, we run the risk of automation   
   complacency where we become even more reliant on the generative AIs outputs   
   and are thrown to the fringes of being in the loop.    
      
   And as more time passes by, this is the moment where the issue of skill   
   erosion might blow to our face: the prompt that produced the accurate output   
   for a long time does not do so any longer as the underlying GenAI tools model   
   changes (such as OpenAI forcing the move from GPT4 models to a single GPT5   
   model) or the software that builds upon the lines of code created with GenAI   
   tools breaks.    
      
   If we have become complacent about conducting work mindfully and digging into   
   the details, it is likely that skills have eroded over time.    
      
   Therefore, on an individual level, critical thinking and maintaining active   
   reflection is essential as GenAI tools responses can at first look    
   convincing, but as we know, looks can be deceiving as the responses can be   
   suboptimal or partly hallucinated.    
      
   This issue is even more profound to junior employees who will likely have    
   less chances to immerse themselves with the work context and facing less   
   challenges to solve if they are mostly evaluating GenAI tools outputs  in   
   other words, there are less chances to learn on the job to become    
   experienced.   
      
   Given the urgency and the clear risks associated with the   
   phenomenon of skill erosion, why isn't this issue pushed atop agendas?    
      
    EP: What makes this phenomenon tricky is its latent nature. If a company   
   fully automates a business process, there are no problems as long as the   
   system works.    
      
   This was true in our case organization as well.    
      
   The system was in a way too perfect, effectively optimizing the client   
   companies optimization of their fixed assets. So why push something atop   
   agendas if there are no problems?    
      
   Problems arise then when something goes wrong. In this case, it was a top   
   management decision to discontinue the automation system.    
      
   The environment changed, leading to the discovery of the detrimental latent   
   effects on employee skills. In some other contexts, it might be some other   
   form of trigger that unearths the long-term impactful problems related to   
   automation reliance.   
      
   Anything else you want to discuss that wasn't covered in the questions above?   
      
    EP: Which skills are such that should be retained and which skills can be   
   forgotten or eroded? Drawing this line seems to be problematic.    
      
   Partly due to the changing environments. Something considered redundant now   
   might not be considered redundant in the future.    
      
   We encourage companies to engage in scenario analysis, what are the possible   
   and foreseeable alternative scenarios on organizational, technological, and   
   environmental fronts?    
      
   How likely is it that an automation technology or an AI-tool that an   
   organization has deployed suddenly becomes unavailable?    
      
   How likely is it that an environmental change impacts the required necessary   
   skills in the business process that I am personally responsible?    
      
   What if our organization makes a strategic decision that impacts our IT   
   infrastructure in a way that jeopardizes our IT?    
      
   These are the questions that we would like companies to consider.   
   ======================================================================   
   Link to news story:   
   https://www.techradar.com/pro/researchers-warn-that-skill-erosion-caused-by-ai   
   -could-have-a-devastating-and-lasting-impact-on-businesses-but-it-may-already-   
   be-too-late   
   $$   
   --- SBBSecho 3.28-Linux   
    * Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (1:2320/105)   
   SEEN-BY: 105/81 106/201 128/187 129/14 305 153/7715 154/110 218/700   
   SEEN-BY: 226/30 227/114 229/110 111 206 300 307 317 400 426 428 470   
   SEEN-BY: 229/664 700 705 266/512 291/111 320/219 322/757 342/200 396/45   
   SEEN-BY: 460/58 712/848 902/26 2320/0 105 304 3634/12 5075/35   
   PATH: 2320/105 229/426