{"course":{"slug":"ai-actually-explained","title":"AI, Actually Explained","url":"https://addaly.com/learn/ai-actually-explained","level":"beginner","track":"start"},"lesson":{"slug":"what-to-do-about-it","title":"What an early-career person can actually do","url":"https://addaly.com/learn/ai-actually-explained/what-to-do-about-it","position":82,"of":92,"minutes":9,"answer":"The evidence points at five moves (learn to check before you can be replaced at producing, go deep in a domain rather than wide in tools, use the tools harder than your peers.","keyIdea":"The evidence points at five moves (learn to check before you can be replaced at producing, go deep in a domain rather than wide in tools, use the tools harder than your peers, aim for the role with a name on it, and keep the skills the tool does not touch) and the honest caveat is that no one has done this before."},"body":{"format":"markdown","text":"## First, the caveat\n\nNobody has been early in a career through this before. The advice below is what the evidence in this module supports, and the evidence is three years old. Treat it as the best available reading of an unfinished experiment, not as a map.\n\n## The five moves\n\n**Get to judgement faster than the last generation had to.** The ladder lesson showed the rung going missing: the routine task that trained juniors is the model's now. The reused lesson at the start of this module said it plainly: learn to check work rather than only produce it. Concretely: whenever you use a tool, verify the specific before you pass it on, and keep a note of where it was wrong. Within months you have a map of the frontier for your trade that most seniors do not have, because they learned the trade before the tool existed. That map is a skill and a scarce one.\n\n**Go deep in a domain rather than wide in tools.** The lesson on who gains showed the model supplies the median of general practice. What it does not have is your client's history, your industry's unwritten rules, your region's regulations, the reason the standard answer is wrong here. That is domain knowledge, and it is what the expert developers had that the tool lacked. Tool skills are worth having and they are worth less each year, because the tools get easier. The domain does not get easier.\n\n**Use the tools harder than your peers.** Every experiment in this module found that the people using the tools well outperformed the people who did not, by a large margin on the right tasks. The freelance lesson found buyers already assume you use them. The person who is in the trade *and* fluent with the tools is the one the studies say does best; the person in the trade who refuses them is competing with the median at a handicap.\n\n**Aim for the role with a name on it.** Across all six trades the value moved toward accountability: the signature, the licence, the person who carries the consequence. Those roles are structurally protected, because a model cannot be struck off, sued or sacked. Professions with a licence (law, medicine, accountancy, engineering) hold their ladders better for this reason. In trades without one, the equivalent is reputation: being the person a client trusts by name.\n\n**Keep the skills the tool does not touch.** Physical, situated, relational work is the least exposed, and the earlier lesson on judgement showed why doing a task yourself sometimes is the only way to stay able to. Choose the tasks you want to stay good at and do them without the tool, deliberately.\n\n## Two things not to do\n\nDo not learn \"prompt engineering\" as a career. In 2023 there were postings for the role at high salaries; by 2025 it had largely dissolved into ordinary use, as the models got better at understanding plain requests. The skill of giving a model good material, from the module on putting it to work, is real and is learned in an afternoon.\n\nDo not wait to see how it settles. The freelance data moved within months, the payroll data within two years, and neither waited for anyone's policy. The adjustment is under way and the people who did well in every automation of the past were the ones who moved before the shape was clear.\n\n## The employer's side, for completeness\n\nYou will also be on the other side of this. When a firm you work for decides to cut the junior rung, the ladder lesson is the argument against; when it announces that a tool has replaced a team, the support-company reversal is the argument for care. A forecast from the World Economic Forum in 2025, from surveying employers, expected far more roles created than displaced by 2030, but it is a survey of what employers said they expected, and the module has shown you what surveys are worth against stopwatches.\n\n## Where this course leaves you\n\nYou can now read a claim about AI and work with the right questions: task or job, exposed or replaced, who gained, where on the frontier, how far demand stretches, whose labour is hidden. The catalogue's course on AI careers takes the practical side further. The final module of this course turns to the larger question underneath all of it: who owns the thing, who is writing the rules, and what nobody has settled yet."},"check":{"question":"Why does this module advise depth in a domain over breadth across AI tools?","options":[{"text":"Because the tools are too hard to learn without formal training, so time spent on them is wasted for most people.","correct":false,"tempting":"Reverses the point; the tools get easier every year, which is why skill in them depreciates."},{"text":"Because the model supplies the median of general practice, and lacks the specific context a domain expert holds.","correct":true},{"text":"Because employers do not yet value AI skills.","correct":false,"tempting":"Contradicts the evidence; the people using the tools well outperformed those who did not in every study."},{"text":"Because domain knowledge is easier to acquire than tool knowledge.","correct":false,"tempting":"Gets the difficulty backwards; domain depth takes years, and the point is that this is why it holds value."}]},"prev":{"slug":"six-trades-honestly","title":"Six trades, as they stood in 2025","url":"https://addaly.com/learn/ai-actually-explained/six-trades-honestly"},"next":{"slug":"the-stack-and-its-chokepoints","title":"Four layers and their chokepoints","url":"https://addaly.com/learn/ai-actually-explained/the-stack-and-its-chokepoints"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'What an early-career person can actually do', in AI, Actually Explained. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/ai-actually-explained/what-to-do-about-it (Accessed: 13 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"What an early-career person can actually do","courseTitle":"AI, Actually Explained","module":{"index":8,"slug":"ai-and-jobs","title":"AI and work, with the evidence"},"answer":null,"keyIdea":"The evidence points at five moves (learn to check before you can be replaced at producing, go deep in a domain rather than wide in tools, use the tools harder than your peers, aim for the role with a name on it, and keep the skills the tool does not touch) and the honest caveat is that no one has done this before.","hero":{"kind":"plate","image":null,"alt":null,"caption":null},"blocks":[{"t":"h2","id":"first-the-caveat","runs":[{"text":"First, the caveat"}]},{"t":"p","runs":[{"text":"Nobody has been early in a career through this before. The advice below is what the evidence in this module supports, and the evidence is three years old. Treat it as the best available reading of an unfinished experiment, not as a map."}]},{"t":"h2","id":"the-five-moves","runs":[{"text":"The five moves"}]},{"t":"p","runs":[{"text":"Get to judgement faster than the last generation had to.","b":true},{"text":" The ladder lesson showed the rung going missing: the routine task that trained juniors is the model’s now. The reused lesson at the start of this module said it plainly: learn to check work rather than only produce it. Concretely: whenever you use a tool, verify the specific before you pass it on, and keep a note of where it was wrong. Within months you have a map of the frontier for your trade that most seniors do not have, because they learned the trade before the tool existed. That map is a skill and a scarce one."}]},{"t":"p","runs":[{"text":"Go deep in a domain rather than wide in tools.","b":true},{"text":" The lesson on who gains showed the model supplies the median of general practice. What it does not have is your client’s history, your industry’s unwritten rules, your region’s regulations, the reason the standard answer is wrong here. That is domain knowledge, and it is what the expert developers had that the tool lacked. Tool skills are worth having and they are worth less each year, because the tools get easier. The domain does not get easier."}]},{"t":"p","runs":[{"text":"Use the tools harder than your peers.","b":true},{"text":" Every experiment in this module found that the people using the tools well outperformed the people who did not, by a large margin on the right tasks. The freelance lesson found buyers already assume you use them. The person who is in the trade "},{"text":"and","i":true},{"text":" fluent with the tools is the one the studies say does best; the person in the trade who refuses them is competing with the median at a handicap."}]},{"t":"p","runs":[{"text":"Aim for the role with a name on it.","b":true},{"text":" Across all six trades the value moved toward accountability: the signature, the licence, the person who carries the consequence. Those roles are structurally protected, because a model cannot be struck off, sued or sacked. Professions with a licence (law, medicine, accountancy, engineering) hold their ladders better for this reason. In trades without one, the equivalent is reputation: being the person a client trusts by name."}]},{"t":"p","runs":[{"text":"Keep the skills the tool does not touch.","b":true},{"text":" Physical, situated, relational work is the least exposed, and the earlier lesson on judgement showed why doing a task yourself sometimes is the only way to stay able to. Choose the tasks you want to stay good at and do them without the tool, deliberately."}]},{"t":"h2","id":"two-things-not-to-do","runs":[{"text":"Two things not to do"}]},{"t":"p","runs":[{"text":"Do not learn “prompt engineering” as a career. In 2023 there were postings for the role at high salaries; by 2025 it had largely dissolved into ordinary use, as the models got better at understanding plain requests. The skill of giving a model good material, from the module on putting it to work, is real and is learned in an afternoon."}]},{"t":"p","runs":[{"text":"Do not wait to see how it settles. The freelance data moved within months, the payroll data within two years, and neither waited for anyone’s policy. The adjustment is under way and the people who did well in every automation of the past were the ones who moved before the shape was clear."}]},{"t":"h2","id":"the-employer-s-side-for-completeness","runs":[{"text":"The employer’s side, for completeness"}]},{"t":"p","runs":[{"text":"You will also be on the other side of this. When a firm you work for decides to cut the junior rung, the ladder lesson is the argument against; when it announces that a tool has replaced a team, the support-company reversal is the argument for care. A forecast from the World Economic Forum in 2025, from surveying employers, expected far more roles created than displaced by 2030, but it is a survey of what employers said they expected, and the module has shown you what surveys are worth against stopwatches."}]},{"t":"h2","id":"where-this-course-leaves-you","runs":[{"text":"Where this course leaves you"}]},{"t":"p","runs":[{"text":"You can now read a claim about AI and work with the right questions: task or job, exposed or replaced, who gained, where on the frontier, how far demand stretches, whose labour is hidden. The catalogue’s course on AI careers takes the practical side further. The final module of this course turns to the larger question underneath all of it: who owns the thing, who is writing the rules, and what nobody has settled yet."}]}],"check":{"question":[{"text":"Why does this module advise depth in a domain over breadth across AI tools?"}],"options":[{"text":[{"text":"Because the tools are too hard to learn without formal training, so time spent on them is wasted for most people."}],"correct":false,"tempting":[{"text":"Reverses the point; the tools get easier every year, which is why skill in them depreciates."}]},{"text":[{"text":"Because the model supplies the median of general practice, and lacks the specific context a domain expert holds."}],"correct":true,"tempting":null},{"text":[{"text":"Because employers do not yet value AI skills."}],"correct":false,"tempting":[{"text":"Contradicts the evidence; the people using the tools well outperformed those who did not in every study."}]},{"text":[{"text":"Because domain knowledge is easier to acquire than tool knowledge."}],"correct":false,"tempting":[{"text":"Gets the difficulty backwards; domain depth takes years, and the point is that this is why it holds value."}]}],"order":[3,1,2,0],"why":null},"prev":{"slug":"six-trades-honestly","title":"Six trades, as they stood in 2025","minutes":10},"next":{"slug":"the-stack-and-its-chokepoints","title":"Four layers and their chokepoints","minutes":9}}}