In partnership with
I have been watching the AI job conversation move in cycles for about two years. A report drops. Someone argues it is overblown. Another report drops. The same argument restarts. Both sides find the numbers they need.
What gets lost in that loop is the specific, ground level picture. What is actually happening right now to real work and real income. Not predictions spread across decades. Not projections tied to conditions that may never fully arrive.
I looked at it properly. The picture is less dramatic than the loudest voices suggest and more serious than the skeptics want to admit.
The Question Itself Is Framed BadlyThe standard framing of this conversation is a binary. Either AI is coming for all jobs, or it is mostly noise and things will sort themselves out. Neither of those is accurate. Neither is useful to anyone trying to understand their actual situation. What is actually happening is more specific. Certain tasks inside certain roles are being automated. Not the whole role, not the whole industry, but enough of the work to change what companies need to hire for. That is a meaningful distinction. A law firm does not need to eliminate its legal team. It might need fewer junior associates if one experienced lawyer with good tools can cover the research that three people used to handle. That is how the compression works. Not replacement in the dramatic, science fiction sense. Reduction at the edges of knowledge work teams, in the most repeatable parts of those roles. |
Numbers Worth Paying Attention ToThe large projections get most of the attention. Goldman Sachs projects that 6 to 7 percent of the US workforce could be displaced if AI adoption reaches full potential. The World Economic Forum puts the global figure at 92 million roles by 2030. These numbers matter structurally, but they are spread across decades and include roles that would have shifted for other reasons anyway. The ground level numbers are more telling. In the first six months of 2025 alone, around 78,000 tech jobs were cut and companies directly linked those cuts to AI adoption. Over the full year, roughly 55,000 job losses were formally attributed to AI by employers out of 1.17 million total layoffs, which was the highest layoff volume since 2020. The Anthropic labor study published in February 2026 identified computer programmers, customer service representatives, and financial analysts as among the most exposed roles. Not because AI can replace those jobs entirely, but because it can handle a high proportion of the repeatable tasks inside them. That is the mechanism worth understanding. |
What Is Going FirstThe pattern is consistent across industries. Entry level positions in white collar work are the most exposed right now. Graduate hiring pipelines, junior analyst cohorts, entry level developer roles. The reasoning is straightforward: one experienced person with good AI tools can now produce what three people used to produce. The bottom of those hiring structures is thinning out. Wall Street banks are planning to cut around 200,000 back office roles over the next three to five years. A King's College study found that higher-paying firms lost around 9.4 percent of jobs between 2021 and 2025, a window that maps almost exactly onto the release and widespread adoption of ChatGPT. Bookkeeping, routine data analysis, standard legal document review, templated financial reporting. These are the early areas. Construction, agriculture, personal care, and skilled trades remain insulated for now. The physical and relational demands of that work create barriers that current AI cannot cross. The conversation about job displacement is mostly happening among people in knowledge work. That is not a coincidence. Knowledge work is where the real pressure is already measurable. |
Where the Money Is Sitting NowThe income picture is not uniform, which is the part that tends to get simplified in the coverage. Some income streams are compressing. AI assisted copywriting and generic content work has seen rates fall by around 33 percent over the past two years. Standard SEO writing, templated reporting, basic design production. The income in those areas moved downward because the supply of output went up and the cost of producing it dropped sharply. On the other side, AI specific freelance work is expanding. Freelance income in AI work averages between $75 and $300 per hour in major markets depending on specialisation. People working in LLM integration, RAG system development, and AI agent configuration sit at the top of that range. ZipRecruiter data puts the average annual income for AI freelancers at around $99,000, with top earners reaching $200,000. The money did not disappear. It redistributed. The compressing end is where work was already somewhat generic, already somewhat interchangeable. The expanding end is where genuine technical specificity now has a real market that did not exist two years ago. These two things are happening at the same time, in the same economy, which is why the aggregate picture looks confusing from a distance. |
The Honest Version of This StoryAI is not taking over all jobs. It is accelerating a kind of sorting that was already underway. Work that was always somewhat replicable, always somewhat interchangeable, was already under pressure from outsourcing and globalisation. AI moved that process faster and brought it into knowledge work in ways those earlier forces could not reach. The 30 percent of US companies that have already replaced workers with AI did not do it out of ideology. They did it because the economics became straightforward. Nearly 40 percent of companies that adopt AI choose full automation rather than using it alongside existing workers. About one in six employers expects to reduce headcount specifically because of AI in 2026. These are not projections. They are present-tense numbers. The income is sitting where it has always sat in periods like this. With the work that cannot be easily replicated and with the people who genuinely understand the thing that is changing. Not because they saw it coming years in advance or made a calculated bet. Just because they were paying attention to what was actually happening, at the ground level, while the larger argument was still going in circles. |
Email Still Wins. Here's How to Use It Better.
59% of Americans say most marketing emails offer no real value. That's not a threat, it's an opening. Get the AI-powered playbook for building email campaigns that actually convert.
Inside you'll discover:
How top brands achieve 3,600% ROI from email marketing
AI personalization techniques that drive 82% higher conversion rates
Tactics that have delivered 30% better open rates and 50% higher clickthroughs
How to build sequences for every stage of the customer journey, from welcome to re-engagement
Download your free AI-powered email marketing playbook today.