How AI is transforming entry-level jobs

You explored the use of applications like ChatGPT in knowledge work, taking the example of management consultants: How do entry-level professionals use AI tools?

Anne-Sophie Mayer: Junior consultants use ChatGPT and other tools like Copilot and Midjourney for routine tasks and to improve and personalize outputs like presentations and visualizations. In addition, many entry-level professionals use these tools for self-mentoring – for example, to prepare salary negotiations or performance reviews with their managers. And finally, these tools are increasingly becoming members of the team, putting professionals’ ideas to the test, offering opinions and recommendations, and giving brainstorming inputs.

In view of this wide range of possible applications, it was very exciting to observe what implications they’re having on the skills development and career progression of entry-level professionals.

What does the use of AI mean for entry-level professionals’ skills acquisition and development?

First of all, the AI tools allow junior consultants to work much more independently. While juniors traditionally had to solicit the help of experienced colleagues or managers for certain questions or problems, they can now answer many of these questions with their own ‘personal assistant’. As a consequence, however, managers are finding it increasingly difficult to identify knowledge gaps and upskilling needs in new team members. And conversely, junior professionals used to actively learn from these interactions, and these opportunities are disappearing or at least becoming much rarer.

»Individual mentoring will not be so easily replaced by technology: Having someone who recognizes your potential, introduces you to the right people, and brings you on board a project is vital for every career – and only people can do that.«

Anne-Sophie Mayer

So AI is no substitution for learning from the experience of others?

Individual mentoring will not be so easily replaced by technology: Having someone who recognizes your potential, introduces you to the right people, and brings you on board a project is vital for every career – and only people can do that.

Moreover, social networks are not only important for skills acquisition and promotion opportunities, but also for corporate culture itself. Social interactions are the lifeblood of companies, where important information is relayed and knowledge is generated. If these interactions are limited, because people are using more technology for knowledge and communication, this causes problems in information and knowledge diffusion. You see, every AI assistant works with the respective user in isolation, meaning that questions, ideas, and possible solutions remain in the AI tool, instead of being shared with other members of staff and spread throughout the company. This is an aspect that companies have often neglected to consider.

»At the individual level, users can become more creative thanks to AI tools. But at the collective level, we often see the homogenization of content. This puts entry-level professionals under more pressure to signal their added value.«

Anne-Sophie Mayer

And how does work output change as a result of using AI tools? Does everyone become equally good or bad?

At the individual level, users can become more creative thanks to AI tools. But at the collective level, we often see the homogenization of content – an observation that is also confirmed by other studies. This puts entry-level professionals under more pressure to signal their added value. Everybody knows a colleague whose work is mostly copying-and-pasting, but is so silver-tongued that people don’t notice where the person gets his results from – and whether they are even accurate. As such, validating the outputs of junior professionals is becoming an increasingly important part of managerial roles.

So juniors increasingly promote themselves and their value towards their managers by for instance showing them a useful prompt or developing a new LLM that can benefit the department. Thus, self-promotion becomes a new important skill to emphasize one’s added value and standing out from colleagues and the AI tool. While previously it was common to assume that my work speaks for itself, we now see that professionals can no longer take this assumption for granted.

How are managers responding to this development?

Their expectations of entry-level professionals are much higher. Managers often do not tolerate errors anymore. For example, spelling mistakes are no longer accepted; linguistic perfection is expected even of non-native-speakers.

But at the same time, it’s a problem for managers when they cannot tell where the output of junior professionals comes from. How can they then identify the high performers? Who even fits this description nowadays? Is it those who are good at prompting? Or those who don’t use the tools because they don’t need them?

Another danger is that managers receive documents that look great but contain errors or hallucinations. Managers simply don’t have the time to check the factual accuracy of everything their team delivers to them. Yet it would naturally cause great reputational damage if, for instance, incorrect figures are stated in a client presentation.

And how do organizations respond to these challenges?

The consulting firm at which we carried out our study has since introduced interim checks, where junior consultants account for how they arrived at their results and explain the process to their managers. This has created an incentive for them to not just copy and paste outputs. The company has also set up a new training program to teach entry-level professionals basic skills in spite of AI and to show them how to work with AI tools in a sensible, thoughtful manner. In addition, the company tried to optimize their own LLMs to increase the quality of their contents and ensure their verifiability – for example, through integrated commands to always provide sources for statements and show how they reached a result instead of just delivering it.

»But although juniors are increasingly performing more senior tasks, they’re not automatically promoted sooner or paid better. And that raises of course dissatisfaction and concerns.«

Anne-Sophie Mayer

If entry-level professionals deliver better work with the help of AI, do they then rise faster through the ranks?

At the moment, there’s a paradoxical situation where junior consultants are doing senior work at an earlier stage and for instance manage client projects already in their first or second year. But although juniors are increasingly performing more senior tasks, they’re not automatically promoted sooner or paid better. And that raises of course dissatisfaction and concerns.

Do you have any advice for entry-level professionals?

Business students often ask me this question in my lectures. Many of my master’s students would like to enter management consultancy and are watching current trends with concern.

Naturally, not using the AI tools is not a solution, as they can offer a lot of added value, both for the individual and the organization. But it’s becoming increasingly important to understand how the tools work and how to work with them in effective and innovative ways. Then entry-level professionals can make a very valuable contribution to companies by reporting on their experiences, making others aware of risks, and highlighting best practices.

So I recommend that people acquire technical skills – such as “How do I provide the right prompts?” or “How do I build an LLM?” – but also focus more on their social networks: Even if you can work more independently thanks to AI, social relationships within the department and the company are more important than ever for building expertise, gaining access to exciting projects and roles, and for personal and professional development.

Anne-Sophie Mayer is Professor of Digital Work at LMU Munich School of Management.

Publication:

Anne-Sophie Mayer, Reza M. Baygi, Reinout Buwalda: Generation AI: Job Crafting by Entry-Level Professionals in the Age of Generative AI. In: Business & Information Systems Engineering 2025

More rethinking in this issue:

• The Interpreter: Jelena Spanjol on disruptive innovation. Read now!

• RNA – so much more: It was long thought that RNA merely mediates protein synthesis. But by now it is abundantly clear: these molecules perform many more tasks in the body. Read now!

• Tailored cancer therapy: Marion Subklewe and Sebastian Kobold are using the immune system to target tumors, opening up new perspectives for patient-specific treatments. Read now!

Read more articles from LMU's research magazine in the online section and never miss an issue by activating the magazine alert!