AI will not hit every part of a company equally hard. It is easy to talk about automation as if it affects some anonymous group far away. Factory workers. Customer service. Junior developers. Someone else. But I believe some of the most dangerous places to be right now are much closer to the management team than that. HR, marketing and finance are three functions where a great deal of work has long been able to hide behind meetings, reports, presentations and processes.
That does not mean these functions are unimportant. A company without functioning recruitment, marketing and financial control quickly becomes a bad company. The problem is that internal fog often builds up precisely in important functions. When something sounds strategic, business critical or culturally important, it becomes easier to create work that looks professional without anyone really asking what it leads to.
AI makes this uncomfortable because it removes much of the fog
What used to take two days can sometimes be done in twenty minutes. A report can be summarized. A draft can be written. A campaign idea can be produced. A brief can be structured. A presentation can be built. An analysis can be started. That does not mean the result automatically becomes good. It is often quite mediocre if no one with judgment directs the work. But it does mean that many tasks can no longer be defended simply because they take time. That is a brutal change.
In many organizations, time has long been confused with value. If something has taken a long time, it has felt important. If many people have been involved, it has felt serious. If it has required several meetings, a document, a presentation and a round of alignment, it has looked like management.
HR is a good example. Real HR can be enormously valuable. Helping the company recruit better, develop managers, remove bad behavior, build functioning structures and make sure people do not break is genuinely important. But a lot of HR work is not that. Much of it consists of policies, internal programs, values exercises, onboarding material, employee surveys and initiatives that never change how managers make decisions.
AI will not replace good HR. AI can, however, replace large amounts of HR material.
That should worry every HR department that has measured its value by how much it produces rather than by how much better the organization becomes. If HR does not make the company better at finding, developing, retaining and sometimes exiting the right people, what is the point? A policy is not culture. A training program is not management. An employee survey is not leadership. They are only tools, and tools are not very impressive if they are not used for anything.
Marketing may be even more exposed. Few areas have produced so much material with such a weak connection to results. Campaign plans, content calendars, brand platforms, tone of voice documents, social media posts, analyses, target group descriptions and presentations about the customer journey. All of it can be relevant. But only if it leads to more people understanding, trusting and buying from the company.
That is where AI becomes dangerous. AI can produce almost unlimited amounts of marketing material. It can write a hundred posts, ten campaign concepts and fifty headlines before lunch. If the value of the marketing department lies in producing text, ideas and presentations, it is no longer a very safe place. The question quickly becomes why many people should be needed to create something a tool can create in a first version.
The marketer who becomes more valuable because of AI is the one who can see what actually sells. The one who dares to choose what not to do. The one who can formulate something sharp instead of something everyone can approve. The one who can use AI to test more, learn faster and spend their own time on judgment. But the person who only produces more content has a problem. More content is not the same as more marketing.
Finance may seem safer because the numbers must always be right. But the same problem exists there too. Finance is vital in a company. Cash flow, margins, invoicing, taxes, salaries and control must work. That is not fluff. It is the oxygen. But finance functions can also grow into something that mostly produces supporting documents, budget versions, reporting packages and internal controls that do not lead to better decisions.
AI will not remove the need for financial responsibility. It will, however, make it harder to justify every manual compilation, every report no one reads and every process that mostly exists because it has always existed. The important finance person of the future is not the one who can move numbers between systems. It is the one who can see what the numbers mean, detect risks early and help the company make better decisions.
This is not an AI question, it is a value question
AI simply makes the question impossible to avoid. What does this function contribute? What does this person contribute? What becomes better for the customer, the company or the employees when this work is done? If the answer is unclear, AI will make that lack of clarity much harder to defend.
It is also a management question. Bad managers will use AI as an excuse to cut people without understanding what they are actually removing. That is dangerous. There is an enormous amount of tacit knowledge in organizations. People who understand the customers, the history, the systems, the relationships and the risks. If you cut them away too quickly, you can end up with polished AI generated documents and a company that works worse.
But the opposite is just as dangerous. Pretending that nothing has changed. Continuing to staff as before, meet as before, report as before and produce internal documents as before, even though the tools have changed the conditions. That is not considerate. It is cowardly. When the job changes, the manager has to say so.
The reasonable path is clarity. HR must be able to show how it makes managers better and recruitment more accurate. Marketing must be able to show how it creates demand, trust and business. Finance must be able to show how it creates control, foresight and better decisions. Everything else needs to be questioned.
That does not mean everyone has to become a technician. It means everyone has to become better at their real job. AI takes over parts of production. That makes the human contribution more visible. Judgment. Prioritization. Responsibility. Customer understanding. Business understanding. The courage to say no. The ability to see what actually matters.
That is why HR, marketing and finance are dangerous places to hide. Not because the functions lack value, but because the value has to become clearer. Those who have hidden behind processes, documents and meetings will have a hard time. Those who actually make the company better will become even more important.
Examples of companies already cutting in these departments
| Company name | Number of people | Source |
|---|---|---|
| Amazon | Up to around 1,500 in HR/PXT, based on reports of up to 15 percent of a PXT organization of more than 10,000 people. Amazon has not confirmed the plan. | Reuters and HR Grapevine |
| Uber | 23 percent of the HR and recruiting team. Exact number not disclosed, but less than 1 percent of a global workforce of around 34,000 people. | San Francisco Chronicle |
| Disney | Around 1,000 in total, with clear impact on the marketing organization after consolidation of marketing functions. | Reuters and Business Insider |
| Salesforce | Fewer than 1,000, including roles in marketing, product management, data analysis and Agentforce. | Business Insider and Salesforce Ben |
| Salesforce | 86 in a later round, linked to Agentforce, MuleSoft and Marketing Cloud, with roles including sales, general administration and tech/product. | Business Insider and Salesforce Ben |
| Fewer than 200 in total in a round that affected HR, ad sales and cloud, among other areas. | LinkedIn News summary of CNBC, Business Insider and Bloomberg | |
| Hundreds in global recruiting in 2023 when the company slowed hiring. | CRN and Silicon UK | |
| Meta | 10,000 in total in the 2023 round, with the first wave aimed at the recruiting team. | Axios |
| Meta | 4,000 in April 2023 after a smaller earlier round targeting the recruiting team. | Reuters |
| Omnicom | More than 4,000 after the acquisition of Interpublic. Relevant to the marketing and agency side, but not an internal marketing department. | Intellizence |
| Kuda Bank | 19 marketing roles, nearly half of a 40 person marketing team, according to reporting on a broader restructuring. | Technext |
| Disney | Around 1,000 in total, with reporting also mentioning marketing, publicity, casting, development and corporate finance. | HRKatha and Reuters |
