Artificial Idea | AI careers · practical prompts · no hype Monday, August 11, 2025 · Issue #3 · Jobs
This week's breakdown
The conversation about AI and employment tends to flatten everything into a single, undifferentiated anxiety. "Jobs are at risk." Which jobs? All of them, apparently. The radiologist and the receptionist. The software engineer and the social worker. The consultant and the cashier. When everything is at risk, the warning becomes useless , because it tells you nothing about where to stand, what to do, or how urgently to do it.
The data, when you actually read it rather than skim the summary, tells a more structured story. AI is not advancing uniformly across the labour market. It is advancing along specific lines , targeting specific characteristics of work, leaving others largely untouched, and creating new demand in places most people are not looking. Understanding that structure is not an academic exercise. It is the most practical thing a professional can do right now, because it tells you exactly where your role sits on the map and what, if anything, you need to change about it.
Here is what the structure actually looks like.
The 3 types of jobs being displaced
Type 1: Single-task roles
These are roles whose primary function is the repeated execution of one well-defined task. Data entry operators, basic transcriptionists, certain categories of customer service agents handling tier-one queries, document formatters, simple invoice processors. The defining characteristic is that the work is high-volume, low-variability, and can be described in a precise ruleset.
AI does not get tired. It does not have off days. It does not need a salary, benefits, or a desk. For work that fits this description, the economic case for automation is overwhelming and the transition is already largely complete in organisations that were early adopters. If your role is primarily defined by one task that follows a consistent pattern, the displacement risk is not coming , it is here.
Type 2: First-draft and first-pass roles
This category is broader and more relevant to the readership of this newsletter. These are roles where the primary value delivered was the production of an initial output , a first draft, a first-pass analysis, a preliminary research summary , that a more senior person would then review, refine, and act on.
Junior copywriters producing first-draft content. Entry-level analysts producing first-pass market summaries. Paralegals doing initial document review. Junior developers writing boilerplate code. These roles are not disappearing overnight, but they are shrinking , because AI can now produce a credible first draft fast enough that organisations need fewer humans at that stage of the pipeline.
This is where the entry-level squeeze is happening. It is real, it is documented, and it is one of the more serious structural consequences of this technology wave. Junior roles have historically been how people enter industries and build foundational skills. That pipeline is narrowing.
Type 3: Coordination and scheduling roles
A specific category of middle management and administrative work , roles whose primary value was coordinating information between people, scheduling, tracking project status, and producing status reports , is being absorbed by AI tools embedded directly into project management software. When the tool can automatically summarise what happened last week, flag what is behind schedule, and draft the update email, the human layer that existed solely to do those things becomes redundant.
This does not mean all middle management is at risk. It means the middle management roles built primarily on information coordination, rather than judgment, relationship management, or organisational leadership, are under significant pressure.
The 2 types of jobs being created
Type 1: AI-adjacent technical roles
The most visible new job category is the one that gets the most coverage , prompt engineers, AI trainers, model evaluators, machine learning operations specialists, AI safety researchers. These roles are real, they are growing, and they pay well. They are also, for the most part, not accessible without a technical background and are fewer in number than the roles being displaced. This is the category that gets cited whenever someone wants to argue that AI creates as many jobs as it destroys. The argument is not wrong, but it is incomplete, because these roles require skills that the displaced workers often do not have and cannot quickly acquire.
Type 2: AI-augmented professional roles
This is the larger and more consequential category, and the one that gets far less attention. These are not new job titles. They are existing roles , in marketing, finance, consulting, law, education, healthcare, design, sales , that have been fundamentally upgraded by AI fluency.
A marketing manager who can use AI to run ten content experiments simultaneously instead of two is not doing a different job. They are doing the same job at a different level of output and impact. A financial analyst who can use AI to model five scenarios in the time it previously took to model one is not a new kind of worker. They are an existing kind of worker operating at a new level of leverage.
This category is where the majority of the opportunity sits for the majority of professionals reading this. You do not need to retrain entirely. You do not need to become technical. You need to become genuinely fluent in using AI within the context of work you already understand deeply , because deep domain knowledge combined with AI fluency is, right now, a combination that the market is paying a significant premium for.
The WEF estimates that 70% of the skills required for most jobs will change by 2030. That sounds alarming. Read it differently: 30% of what makes you good at your job today will still be what makes you good at your job in 2030, and the other 70% will be augmented rather than replaced , if you engage with the tools rather than wait for clarity that is not coming.
The action
Look at your current role description , the formal one, or the informal version of what you actually do day to day. Categorise your main responsibilities against the framework above. Which parts of your role fall into the three displacement categories? Which parts do not?
The parts that do not are worth investing in. The parts that do are worth learning to augment yourself, before someone else makes that decision for you.
On Thursday we are going deep on the single most powerful prompting technique for professionals , the one that turns AI from a generic assistant into something that thinks, to a useful degree, the way you do. It is simpler than it sounds and most people are not using it.
, The Artificial Idea team
