AI jobs are beginning to change not only the tasks performed by young professionals but also the organisational ladder through which they traditionally advanced. At a Milken Institute Asia Summit panel in Singapore, Goldman Sachs President of Asia Pacific ex-Japan Kevin Sneader said junior recruits were starting their careers by managing artificial intelligence agents rather than spending years performing entry-level analytical work. His warning points to a wider institutional challenge: companies may be adopting new technology faster than they are redesigning careers, training systems and management responsibilities around it.
Sneader described the new model in practical terms. Young employees, he said, now have to manage a “virtual army” of AI agents. That reverses the conventional sequence in many professional services firms, where analysts first spend years running financial models, preparing presentations and completing other routine tasks before moving into supervisory roles. If software performs much of that preparatory work, the traditional path from junior analyst to middle manager becomes less predictable.
The change matters because entry-level work has historically served two purposes at once. It produces immediate output for an organisation, but it also acts as a training ground. New employees learn the firm’s systems, standards and client expectations while gradually taking on more responsibility. The source report does not establish that this model has disappeared, but the comments from Goldman and other panellists indicate that its foundations are being questioned as AI takes over more routine activity.
That creates a problem beyond the possible reduction of particular tasks. If young workers begin by directing AI systems rather than completing the work those systems automate, they may acquire a different mix of skills. They could gain earlier exposure to oversight, judgement and workflow management, while having fewer opportunities to learn through repeated manual analysis. Whether that produces better professionals or leaves gaps in practical understanding is not resolved by the report. What is clear is that the old sequence can no longer be treated as automatic.
The pressure is also moving upwards through organisations. Sneader said the management task was shifting from the middle manager to the front line, adding that Goldman did not yet know what would happen to that group. This is a significant distinction. AI adoption is often described as a question of whether technology will replace workers. The panel discussion presented a more complicated possibility: technology may redistribute responsibility between workers, remove some layers of supervision and require existing managers to perform a different role.
Goldman President John Waldron has described the bank’s operations as a “human assembly line” that is suitable for automation, according to the report. That description frames the issue as an organisational redesign rather than a single software upgrade. Where work is organised as a sequence of repeatable processes, AI can alter the relationship between the people who produce, check and manage those processes. The result may be fewer conventional stages in a career, even where human employees remain central to the operation.
Sandra Peterson, an operating partner at investment firm Clayton, Dubilier & Rice, said the traditional career ladder may not survive. She characterised professional service firms as pyramids in which employees progressed through successive levels if they remained in the system. Her comments underline the structural nature of the shift. A pyramid requires a broad base of junior positions and progressively smaller groups at higher levels. If AI reduces the amount of entry-level work, firms may not simply have fewer tasks; they may have a narrower pipeline through which future specialists and managers are developed.
This raises a difficult question for employers. If fewer graduates are hired to perform analysis and preparation, how will organisations identify and develop the people who eventually need to make complex decisions? The supplied report does not provide an answer, and it does not quantify the number of positions that could disappear or change. But the concern is explicit in the panel discussion: the technology may alter the supply of future managers at the same time as it changes the responsibilities of current ones.
Singapore’s financial regulator is already framing the transition as a workforce-management issue. Chia Der Jiun, managing director of the Monetary Authority of Singapore, said operations staff would need to become managers and supervisors of AI agents. He also said banks would need fewer fresh graduates for analysis and preparatory work. These remarks connect workplace automation to public institutions responsible for skills and economic preparedness. The issue is not limited to what individual companies choose to deploy; it also concerns how education systems, regulators and employers prepare people for changed roles.
Chia said Singapore had been preparing by training bank staff in AI skills and working with universities to ensure graduates were ready for the new environment. The report does not detail the content, scale or outcomes of those initiatives. It does, however, show an institutional response with two parts: retraining people already inside the sector and adjusting the supply of new graduates. Both are necessary if the employment system is to keep pace with changing workplace structures, although the report does not establish whether either approach is sufficient.
The urban dimension of this transition lies in the relationship between professional work, institutions and the cities that concentrate them. Singapore is presented in the report as a major financial centre where banks, regulators and universities are confronting the same technological shift at different points in the workforce pipeline. The city’s response illustrates how changes inside offices can quickly become questions of urban economic resilience: what jobs are available to graduates, what skills employers demand and how institutions absorb workers whose previous career path is narrowing.
For cities with large financial and professional-service sectors, the most immediate challenge is therefore not only job displacement. It is the redesign of the passage from education into employment. The conventional entry-level role provided a recognisable first step, while the middle-management position represented a later stage of progression. The comments at the Singapore panel suggest that AI is placing pressure on both ends at once. Fewer junior roles may be available, while established managers may find that coordination and oversight are increasingly performed by frontline employees working with AI systems.
This creates a potential gap between formal qualifications and workplace readiness. Universities may continue preparing graduates for analytical and preparatory tasks that firms increasingly automate, while employers may expect new recruits to manage systems they have not previously encountered. Chia’s comments show that Singapore’s regulator sees collaboration between banks and universities as part of the response. The source does not indicate whether similar arrangements exist elsewhere, but it establishes the institutional problem clearly.
The future of middle management is equally uncertain. Sneader called the redeployment of existing middle managers a “generational challenge for many, many segments”. That wording matters because it places the issue beyond banking. The report identifies finance as the immediate setting, but the underlying question concerns any sector in which routine work feeds a layered hierarchy. Where automation changes the distribution of tasks, organisations must decide whether managers will supervise people, AI agents, or combined systems, and which responsibilities still require human judgement.
The panel also offered a more positive interpretation for young people. Sneader said there had never been a better time to become an entrepreneur because fewer resources were needed to start a business. That claim was presented as his view, not as an established outcome. It does point to a second pathway emerging alongside the threatened career ladder: AI may reduce the resources required to create a venture, even as it weakens some conventional routes into established firms.
The evidence supplied here does not support a conclusion that mass unemployment is imminent, nor does it show that AI will remove all entry-level or middle-management work. It does support a narrower and more consequential finding: senior industry and regulatory figures are observing changes in the organisation of work, the skills expected from employees and the number of graduates needed for traditional analytical roles. The institutional response is beginning to focus on training, university coordination and the supervision of AI agents.
The larger urban question is whether cities and their employers can redesign career systems as quickly as technology changes job structures. Singapore’s example shows the components of that challenge: companies altering everyday roles, regulators anticipating new operational responsibilities, universities being asked to prepare graduates differently and managers facing uncertain redeployment. What remains unknown is how many roles will change, how quickly firms will act and whether new forms of work will provide an equally reliable route into professional life. Those are the developments that will determine whether AI produces a more capable workforce or breaks the ladder through which that workforce has traditionally been built.
