Artificial Intelligence (AI) is becoming a hiring criterion for certain early-career roles at Swiss bank UBS. AI literacy therefore no longer concerns only technical occupations, but also students, new graduates, interns, and people in traditional knowledge jobs. Experiences from companies and schools also show that frequent use is not the same as effective or responsible use.
What AI literacy means today
AI literacy starts with being able to match a tool to a task and critically review its output. It includes writing a useful request, checking the response against reliable information, and recognizing when human expertise remains necessary. Simply producing as many texts, images, or analyses as possible does not demonstrate competence.
The new UBS hiring requirement makes this shift visible. Graduates and interns joining Global Banking and Markets in 2027 are expected to explain during interviews how they can use AI to improve results and efficiency. According to the report, the requirement will sit alongside a strong academic record and is also expected to apply to other newly advertised positions.
UBS also says AI literacy should complement academic and social abilities rather than replace them. That qualification matters: AI might organize information or prepare an initial draft for a client presentation. Deciding whether the financial analysis is plausible and how it should be explained to clients remains a professional and interpersonal responsibility.
Spanish bank Santander is also seeking “advanced AI users” for some trainee programs, according to documents cited in the supplied report. A pattern is emerging within banking: using AI is moving from optional knowledge into regular job profiles. How employers can assess that skill reliably is still unresolved.
Where AI knowledge is already expected
Banks increasingly automate routine work that has traditionally been assigned to junior employees. The examples include financial analysis, research, and client presentations. UBS is also testing avatars of analysts for presentations, meaning digital representations that can stand in for a person when communicating with clients.
This changes the starting point for applicants. A strong degree will no longer be enough by itself for the specified UBS roles; you will also need to explain how AI improves your work. A concrete example is likely to reveal more than a list of familiar tools, such as describing how you create an initial research overview, verify its sources, and remove inaccurate claims.
Advice from US entrepreneur Mark Cuban points in a similar but broader direction. A report on his recommendations for job seekers says applicants should not focus exclusively on large corporations and should also consider smaller companies, where he believes their chances may be better. The source does not provide specific AI requirements from these companies, however, so this remains personal advice rather than evidence of a general hiring trend.
The development also extends well beyond recruitment. AI is already part of everyday work for many people in information and communications technology (ICT). A Swiss survey of 492 ICT workers found that the tools are used particularly for development, knowledge, text, and data-related tasks.
How AI is changing jobs
The Swiss ICT survey presents a mixed picture. According to the Syndicom and Ecoplan survey, 52 percent of respondents report higher productivity, while 46 percent experience greater work intensity. Another 44 percent are frequently unsure whether AI results contain errors.
Higher productivity and relief from work are therefore not the same outcome. If a draft is produced more quickly, the saved time may be absorbed by additional tasks, tighter deadlines, or more extensive checks. Sixty-eight percent of respondents think more often about how AI will affect their occupation, although the source notes that these thoughts can be either positive or negative.
How employees acquire their skills is also revealing: 87 percent learned to use AI through experimentation at work. Many would nevertheless like their employers to provide workshops or courses, along with clearer information about opportunities, risks, and legal or ethical guidelines. According to the survey, high-level corporate strategies are more visible than concrete changes to workflows and professional development.
A stark example from Nairobi shows how deep the disruption can become. After ChatGPT launched in 2022, the business of writing academic papers for overseas students collapsed. At its peak, researchers estimate that at least 40,000 people in Nairobi worked in this already problematic industry; one long-time writer eventually charged between $40 and $70 per paper.
The report on contract work in Kenya also describes declines in transcription, data annotation, and content moderation. Data annotation is the process of labeling information so it can be used to train AI systems. Some remaining workers became “humanizers” who modify AI-generated text to bypass plagiarism checks, which is hardly a stable or convincing new career model.
For Europe’s banking sector, Morgan Stanley analysts expect more than 200,000 jobs to disappear within five years, according to the UBS report. That is a forecast, not a confirmed outcome. At the same time, JPMorgan’s European chief has warned that junior employees should not lose their grasp of professional fundamentals as work becomes automated.
How schools are responding to the new standard
Schools consequently face a dilemma. They are expected to prepare students for workplaces that demand AI knowledge, while avoiding the premature outsourcing of independent thought, writing, and reasoning to software. New York City is taking a considerably more restrictive approach than the employers described above.
Starting with the new school year, about 600,000 public school students through eighth grade will not be allowed to use AI tools. Under the New York City Department of Education rules, individual screens are prohibited through third grade, while companion chatbots, conversational systems designed to act like personal companions, are banned at every grade level. The department recommends no more than 45 minutes of daily screen time for middle school students.
Teachers may continue using AI to prepare lessons, but not to grade students. Parent groups consider even these restrictions insufficient and are calling for a complete two-year moratorium on classroom AI. A task force of teachers, policymakers, experts, and union representatives is expected to produce a report by April 2027 as the basis for future policy.
This approach contradicts the assumption that earlier and more frequent exposure automatically produces stronger AI literacy. It prioritizes core abilities and an age-dependent introduction instead. The sources provide no equivalent nationwide school policy for Switzerland, but the conflict is relevant there as well: workplace expectations are rising while education systems must decide when AI supports learning and when it replaces the learning process.
Pros and Cons of AI literacy as a standard
Pros:
- Career readiness – When banks and other employers require AI knowledge, clearly demonstrated practical experience can support an application.
- Time savings – In the Swiss ICT survey, AI often reduces the time needed for certain development, knowledge, text, and data tasks.
- Better judgment – People who can verify results and recognize limitations are less dependent on confident but potentially inaccurate answers.
- Broader support – AI literacy can help with research, analysis, and presentation preparation without fully replacing professional or social skills.
Cons:
- Greater pressure – Almost half of the surveyed ICT workers experience higher work intensity even though many report greater productivity.
- Errors and uncertainty – Forty-four percent are frequently unsure whether generated results are incorrect.
- Loss of fundamentals – Junior employees or students who outsource tasks too early may fail to develop essential professional and cognitive skills.
- Poor incentives – Measuring only the volume of AI use can encourage employees to use it even when it adds no value.
Meta provides a particularly clear example of the final problem. The company no longer includes AI dashboards and token counters in performance reviews for its engineers. Tokens are processing units into which an AI system divides its inputs and outputs.
According to an internal Meta memo, the former usage-based measure led to “tokenmaxxing,” with employees consuming large quantities of tokens to improve their position in internal rankings. Quality, speed, and the complexity of work are now supposed to count instead. Meta is also testing an AI agent, software intended to carry out computer tasks independently, but some employees are hesitant to connect the tool to private accounts because of privacy concerns.
What this means for you
As a beginner, you do not need to master as many tools as possible at once. A sensible first step is to choose a limited task whose outcome you can judge yourself, such as outlining a presentation or producing an initial summary of existing documents. This teaches you more than basic operation because you will also see where errors, omissions, and ambiguous statements appear.
Step 1: Choose a result you can verify
- Select a recurring text, research, or data task from your daily work or studies.
- Define the expected result and the criteria you will use to judge its quality.
- Compare the AI output with the original material and correct any errors yourself.
As an advanced user, you can gain more by measuring the quality of the entire workflow instead of counting requests. Document where the tool saves time, where additional review is necessary, and which information should not be entered. In professional settings, privacy concerns and internal rules should not appear only after a private account has been connected or confidential material submitted.
Step 2: Demonstrate your competence clearly
- In an application, describe the task, your review process, and the result you achieved.
- Include a limitation or error that you identified and corrected.
- Distinguish clearly between the tool’s contribution and your own professional decision.
AI literacy is developing into a workplace standard, but it is not becoming a substitute for education, experience, and judgment. The examples from UBS, Meta, the Swiss ICT sector, Nairobi, and New York show both higher efficiency and the risks of work pressure, displacement, and weakened fundamentals. The main unresolved issue is whether employers and schools will evaluate the quality of AI use or merely count how often the tools are used.
Sources
- Schweizer Großbank UBS macht KI-Kompetenz zur Einstellungsbedingung – Unknown, 2026-09-07
- KI belastet Arbeitnehmende in der ICT-Branche – Unknown, 2026-09-08
- Wie KI eine ganze Branche in Nairobi auslöschte – Unknown, 2026-09-07
- New York City verbannt KI-Tools aus öffentlichen Schulen bis zur achten Klasse – Unknown, 2026-09-07
- Schluss mit Tokenmaxxing: Meta bewertet Ingenieure nicht mehr nach KI-Nutzung – Unknown, 2026-09-08
- Kleine Unternehmen und KI: Diese Vorteile können Jobsuchende laut Mark Cuban jetzt ausnutzen – Unknown, 2026-09-07
Image: cottonbro studio via Pexels