Read the headlines and you might think a chatbot is about to take every job. That fear is easy to catch. The real story about AI jobs is calmer and more useful.

Through The Glass Creatives is an award-winning design and growth agency. It has worked for a decade with brands from Jacob & Co. to Nuvia Dental Implant Center. Its staff includes designers, writers, strategists, and developers. The team works closely with creative teams going through this shift. So this view comes from real work, not guesswork.

Here is the honest answer. Most jobs will not vanish. But most jobs will change. Asking "what is safe" is the wrong question. The better question is this. What changes will make you valuable in the new setup?

What the actual data says

The most-cited study here is the World Economic Forum's Future of Jobs Report 2023, published in May 2023. It surveyed 803 companies that employ 11.3 million workers worldwide. The main finding is clear. About 23% of jobs are set to change by 2027. That means 69 million new roles and 83 million lost roles. The net loss is around 14 million jobs, or about 2% of today's jobs. That is not a disaster. It is also not nothing.

McKinsey Global Institute looked at this too. Its July 2023 report is "Generative AI and the future of work in America." It found that up to 30% of U.S. work hours could be automated by 2030 if adoption speeds up. But the report stressed one point. "Automated" rarely means the job is gone. More often, some tasks get automated. Then the worker moves to higher-value work.

Goldman Sachs ran the numbers in March 2023. It found that 300 million jobs worldwide could be affected by generative AI. Yet two-thirds of today's jobs are only "exposed to some degree of AI automation." They are not wiped out.

Pew Research added a twist in July 2023. Its report is "Which U.S. Workers Are More Exposed to AI on Their Jobs?" The jobs most exposed to AI are not low-wage roles. They are high-wage office jobs in information work, finance, and management. That flips the old story of automation.

Jobs the research suggests are MOST exposed

Across the WEF, McKinsey, Pew, and Stanford AI Index 2024 datasets, the consistently exposed categories include:

Jobs the research suggests are LEAST exposed

The same datasets converge on a few categories where human work remains hard to displace:

What creative teams are seeing in practice

Many agencies began using AI-assisted workflows in late 2024. Some expected pushback. What surprised them was how cleanly teams split into two groups.

Across creative teams, one pattern keeps showing up. Designers and writers who treat AI as a tool get much more done. They use it to draft, refine, and test ideas. They ship more work, and the quality goes up. Their role grows into something close to senior creative direction. Those who skip AI tend to fall behind on output next to their peers. That gap pushes them to learn the tools.

Across the industry, the teams that struggle most share one trait. Their roles have not changed alongside the tools. The issue is not the person's effort. A role built without AI in the loop is simply less competitive than the same role rebuilt with AI as a partner.

So the question is not "will AI take my job?" The real question is this. Are you willing to rebuild your job around AI?

What "safe" actually means in 2025

The research and real industry experience point the same way. A job is meaningfully safer if it has three traits:

1. It requires judgment under ambiguity

Generative models match patterns. They struggle when the inputs are missing or the goal is truly new. WEF 2023 calls this "analytical thinking." It is the top skill employers say is growing in importance.

2. It involves direct relationship management

Some things do not move to AI. Trust, accountability, and long client or patient bonds stay human. Many clients stick with an agency because they know the team. A tool cannot copy that bond.

3. It has a physical or embodied component

Hands matter. AI has none. Trades, hands-on healthcare, and on-site work stay tied to real bodies in real places.

Most jobs have some of these traits. Almost no job has none. To judge your own job, look at your daily work. How much of it needs judgment, relationships, or hands-on effort? How much is simple pattern-matching that a model can do?

The honest take

No one can promise your job is safe. But the data and industry experience point one way. The people most at risk in the next five years will not lose out because AI is smarter or more creative than them. The risk sits with those who do not pick up the new tools. And that is something a person can change.

Here is the good news. Learning these tools is doable. The better news is simple. People who learn them tend to earn more, not less.

Sources

Related reading: How Long Does SEO Take? The Honest Answer Most Agencies Won't Give You · What Jobs in Your Field Will AI Automate First?