Chapter 7: AI Literacy Liza Long and Abraham Romney A first-year student who has only ever talked to a chatbot on Snapchat for fun is, by the end of one semester, teaching her professors how to prompt more effectively. A technical writer who has spent a career translating engineering specifications into user-friendly manuals discovers that her job now includes auditing an AI system’s claims before they reach a customer. Both these cases are actually the same story, and they prompt a simple question: what does it mean to write for a living in the age of artificial intelligence? You will probably work alongside generative AI for the rest of your career, whether or not you ever choose to use it. Colleagues will use it; documents you receive will be shaped by it; employers will expect you to have judgment about it. AI literacy is the foundation of that judgment: knowing what these systems are, what they do well, how they fail, and what responsibilities attach to their use. AI literacy functions as a professional habit of mind: a set of practices for working alongside generative AI tools while keeping your judgment, your voice, and the trust of the people who read what you produce. It has become a permanent, core competency of technical and professional communication.  Because the technology is changing so rapidly, we are taking a high-level approach to AI literacy with a goal of developing AI fluency. This chapter covers the basics of how LLMs work, how to prompt chatbots and evaluate their outputs, how (and when) to disclose and cite AI collaboration, and how to redesign the documents and assessments you produce so that you remain the human in charge. Image by cocoandwifi from Pixabay Pixabay Content License Chapter Learning Objectives By the end of this chapter, you will be able to: Explain why AI literacy has become a core professional competency in technical and workplace writing. Draft an effective prompt using a structured approach, and explain prompting as a form of authorship. Evaluate AI-generated content for accuracy, bias, and hallucination, and describe practices for verifying claims before they reach a reader. Write a clear, professional AI disclosure statement and cite generative AI tools appropriately in APA style. Describe at least two ways that workplace deliverables and assessments can be redesigned to remain meaningful now that AI can produce fluent text instantly. From Producing Text to Directing It Section objectives: distinguish “writing with AI” from “writing for AI,” and explain why problem formulation is becoming a more valuable professional skill than drafting. For most of the history of technical communication, the job was translation: take complex engineering or scientific information and turn it into prose that a human reader could act on. Generative AI has commoditized the easiest part of that job. Any competent chatbot can now draft a passable memo, a boilerplate proposal, or a first-pass procedure. Fluent language has become abundant and cheap. Judgment has become the scarce resource: knowing what to ask for,  whether the answer is any good, and what to do with it next. Some organizations are already redesigning their information systems around this shift, restructuring documentation into tagged, machine-readable components so that AI tools retrieve accurate context, and training writers to think about two audiences at once: the human reader and the algorithm that will eventually process the same content. Practicing AI literacy means understanding where the profession is heading: toward being the person who curates, verifies, and takes responsibility for what a human-AI system produces together, whether or not your own title ever says “data architect.” That shift starts with a skill our education system rarely teaches directly: problem formulation. We have spent generations training students to answer questions. Designing the question is a rarer, harder skill, and it is the one that matters most now. A vague prompt produces the intellectual equivalent of moldy cheese. A clear, constrained, intellectually ambitious question produces something worth reading. The people who get real value out of AI right now share a few traits: they are curious, they have a genuine problem they want to solve, and they are willing to fail and try again. This is also why “prompt engineering” is the wrong name for the skill we are describing here. The phrase signals that this territory belongs to engineers and coders, and that writers and communicators are only visiting this space. In fact, writers and communicators are prompting natives when “English is the hottest new programming language,” as OpenAI founder and Anthropic employee Andrej Karpathy wrote on X in 2023. A prompt is a rhetorical act: it sets an audience, a purpose, a tone, and a set of constraints, and it draws on exactly the skills technical communicators have always practiced. One accessible way to teach this is the Role, Task, Voice, and Context approach: Define who the AI should act as (role) Tell it what it should do (task) Specify the tone it should use (voice) Provide the background information it needs to get the task right (context). Here is an example: Role: Act as an expert Technical Writer and Usability Specialist. Task: Convert a complex, multi-paragraph technical explanation of Git branching into a clear, numbered troubleshooting guide for a beginner software developer. Voice: Use an instructional, clear, and encouraging tone with short sentences and active verbs. Avoid jargon where possible, and define it immediately if necessary. Context: This guide will be included in an open-source documentation wiki for a university computer science club. The readers understand what coding is but are completely new to version control systems like GitHub, so they often get confused when merge conflicts happen. Experience with a tool reduces how much of this scaffolding you need, and it remains a useful way to think rhetorically about any AI interaction, especially early on. Case Study: From Snapchat to AI Superuser Payton Grond arrived in a first-year writing class as a marketing and communications major with, in her own words, “very, very little” experience using AI for anything academic. Her only real exposure had been talking to Snapchat’s built-in chatbot, mostly as a distraction. By the end of the semester, she had moved well past simply using AI in her own writing: she was teaching classmates, and eventually a professor in another class, how to prompt more effectively. Curiosity, humility, and intention made Payton what her instructor came to think of as a “superuser.” She noticed that different AI platforms behaved differently and matched the tool to the task: one for creative brainstorming, one for structure, one for deep research. And she framed her own use of these tools in a way worth quoting directly: “You’re still working if you’re prompting. You’re the one who is ultimately deciding what gets put on the page.” For Payton, prompting was a form of authorship. That reframing mattered later, in a business applications course, when classmates accused her of cheating for using AI on an assignment graded entirely on formatting, with AI-generated content explicitly permitted. Payton and another student stepped in to explain how prompting actually worked, walking their professor through the process. Students, given the right tools and enough trust, can become some of the best AI educators in the room. Payton’s own recommendation, when asked whether AI literacy needed to be taught formally, was blunt: “It should be taught. Especially in high school.” She had felt the whiplash of arriving at college after years of “no AI, ever” rules, only to find every class handling AI differently. Consistency, transparency, and a little room to experiment made the difference for her, and it will make the difference for the students and colleagues you eventually work alongside. Key Term: Problem Formulation: the cognitive ability to identify, analyze, delineate, and structure a real-world problem so that an AI system can effectively solve it Key Term: Superuser. A person who becomes skilled with AI tools through curiosity, willingness to experiment, and a habit of treating AI as a collaborator whose output still requires their own judgment. Evaluating What AI Gives You Section objectives: identify why large language models hallucinate, and practice calibrated trust when deciding whether to accept, verify, or override AI output. Now that we’ve explored how to interact with chatbots, let’s look at how they work so that we can understand the importance of verifying their outputs. A large language model generates text by predicting, one token at a time, what is statistically likely to come next, based on patterns learned from enormous amounts of training data. It operates purely on probability: no worldview, no lived experience, no mechanism for distinguishing a true claim from a fluent one. That is why these tools hallucinate: they confidently generate citations, statistics, and quotations that sound exactly right but were invented on the spot. An internal Microsoft training document put it bluntly: these systems are “built to be persuasive, not truthful” (as cited in Weiss & Metz, 2023). What this means in practical terms is that you must treat every factual claim, source, and statistic an AI tool gives you as a lead to verify before it goes out under your name. Three other failure modes matter for writers. Bias: models reproduce patterns in their training data, including its stereotypes, its overrepresentation of dominant languages and perspectives, and its gaps. Sycophancy: models tend to agree with the framing you give them, which makes them unreliable critics unless you deliberately invite disagreement. Generic register: model output gravitates toward a smooth, elevated, one-size-fits-all professional voice, fluent, but unanchored in any real situation. Learning to hear that voice, and to notice what it papers over, is a core literacy skill. Evaluating AI output well requires what I have come to call calibrated trust: the moment-to-moment skill of knowing when to follow AI output, when to verify it, and when to override it entirely. Calibrated trust is earned through practice, under real stakes, the same way you learn to trust or distrust a new colleague’s judgment. It also requires sustained attention even when a collaboration is going smoothly, because an AI system runs at constant capacity: it works identically at 2 a.m. and 9 a.m., carrying no fatigue and feeling no weight for a bad decision. That weight belongs entirely to the accountable human. Managing your own attention and your own boundaries around AI-assisted work is a skill you will mostly have to teach yourself, and it’s a critical one. Researchers are already reporting a new kind of burnout experienced by frequent LLM users.. This is also where technical communicators are especially vital, as the people who catch what an algorithm misses. AI-crafted misinformation online tends to follow a recognizable template: a hook, a struggle, a heroic turn, a moral, and a prompt to share. Teaching yourself and the people you work with to recognize that template, and to apply a simple verification habit such as stop, investigate the source, find better coverage, and trace the claim to its original context, is now a basic professional literacy. The same discipline that catches manufactured misinformation on social media catches a fabricated statistic in an AI-drafted report before it reaches a client. Key Term: Hallucination. A large language model’s confident generation of false or fabricated information, including nonexistent sources, quotations, or statistics, that sounds plausible because it matches learned patterns of language, regardless of whether it is true. Key Term: Calibrated Trust. The skill of deciding, in the moment, whether to accept, verify, or override AI-generated output; a skill built through repeated practice under real stakes. Disclosure, Citation, and Professional Practice Section objectives: draft a clear AI acknowledgment statement, and cite generative AI tools appropriately using APA style. Every profession that touches AI-generated content is converging on the same expectation: say what you used, how you used it, and why. In the classroom, I ask students to acknowledge their AI use, share the chat itself, evaluate the AI’s output critically, and explain why they made the choices they did. That same structure translates directly into professional practice. A useful acknowledgment statement names the tool, describes specifically how it was used, and explains how the output was verified or revised before it went out under your name. Example acknowledgement statement: “I used [tool] to [specific task]. I verified the output by [method] and revised it to [what you changed and why].” Citation follows the same logic, just formalized. APA style treats a chatbot as a piece of software with an organizational author: you cite the company, the version, and a description of what it is, then quote or paraphrase its output the way you would any other source. A typical APA reference looks like this: OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat, cited in text as (OpenAI, 2023). The same principle applies whatever the tool: name it, date it, and link to the conversation itself if the platform allows sharing, because a link to the actual exchange is the clearest form of transparency you can offer a reader or a colleague. Two habits will keep you out of trouble here. First, verify every AI-recommended source yourself before treating it as real. These tools can generate a completely convincing citation, author, and journal for an article the author never wrote; the only way to catch this is to look it up. Second, keep citation and disclosure distinct: a citation tells your reader where a specific quotation or fact came from, while a disclosure statement tells them how AI shaped your process as a whole. Professional writing increasingly asks for both. Do we need to cite and acknowledge AI use in every instance? Norms are still developing around this question, and the answer will likely depend on the way AI is used. For example, we are not citing the use of spelling and grammar checks, though LLMs can act in this way. Use your best professional judgment. If the LLM is performing a routine templated task, it may not require an acknowledgment statement. Redesigning the Deliverable Section objectives: describe why fluent AI-generated text undermines older forms of assessment, and identify design principles for documents and assignments that still demonstrate human thinking. For more than a century, the polished document functioned as reliable evidence of thinking. If you could produce clear, well-organized prose, the assumption went, you could think clearly. That assumption is now broken: language is cheap and abundant, and fluency alone proves very little about the person who produced it. The traditional essay and the traditional report were already imperfect proxies for thinking, rewarding stamina and formatting compliance as much as insight. AI has simply made that assessment flaw impossible to ignore. Banning AI outright, or policing it with detection software, misses the more useful question. A colleague of mine, reflecting on faculty debates over AI policy, described three postures worth applying to any workplace deliverable: subtract AI where you need to recover genuine human presence, add it where it expands what is possible, and multiply with it where it can genuinely extend your capacity to create and connect. Research on AI use in writing classrooms backs this up directly: misuse tracks assignment design far more closely than it tracks student intent. Generic, one-size-fits-all assignments, the technical-writing equivalent of “summarize this article,” invite automation. Personalized, well-scaffolded tasks grounded in real context earn genuine engagement instead. The deeper redesign question for technical writers is what you are actually trying to communicate and how you are measuring impact. In a workplace, tracking your effort might mean documenting your prompts and your revisions as a routine part of a project file, the same way engineers keep a design history. The goal stays the same: keep the thinking visible, in a world where fluent writing comes cheap and proves less than it used to. Becoming a Professional Cyborg The skills above are practical, but they sit on top of something less comfortable to talk about: sustained AI collaboration changes how your brain works. After spending several intense weeks co-writing a federal grant application with an AI assistant, I found myself groping for vocabulary to describe what I was feeling, and eventually landed on a word: cyborg. But I am not a glamorous science fiction movie cyborg; I’m the ordinary, unglamorous kind: someone whose thinking has become genuinely entangled with a machine’s, while their body, their accountability, and their exhaustion remain entirely human. Two features of that entanglement are worth naming up front. The first is what I call the asymmetry of endurance: the AI runs at constant capacity, with no circadian rhythm and no point of diminishing returns, so any boundary on the work has to come entirely from you. Deciding when to stop is itself a form of cognitive labor. The second is that this vigilance feels like ordinary, even pleasurable, productive work while it is happening; your body sends the bill later. The practical habit worth building here is simple: schedule your own stopping points before you start a high-stakes AI collaboration, and notice when “one more pass” has become less about the work and more about the machine’s bottomless availability. Naming what you are experiencing, openly and as it happens, is most of what protects you. This will be especially important for technical writers, who are increasingly partnering with LLMs to create their work. Use, Refusal, and Responsibility Some professionals and some of your classmates decline to use generative AI, for reasons that deserve respect: environmental costs, labor and copyright concerns, privacy, or a commitment to developing their own unassisted abilities. Declining specific uses for reasons you can explain and justify is itself an exercise of AI literacy. However, especially in technical writing, refusing to understand the technology is not an option. The responsibilities remain either way: recognizing AI-shaped text, evaluating AI-influenced claims, and making policy decisions about tools you can explain. It’s helpful to understand the technology even if you ultimately choose not to use it, because an informed refusal is more defensible than an uninformed one Whatever you decide, three responsibilities are non-negotiable in this course and in professional life: know your context’s rules (each instructor, workplace, and publication sets its own); disclose use where disclosure is expected; and own the output, every word, as if you had written it, because ethically you have. Chapter Summary AI literacy in technical and professional communication is a bundle of skills: formulating problems clearly enough to prompt well, evaluating AI output with calibrated skepticism, disclosing and citing that collaboration honestly, and redesigning the documents and assessments you produce so they still demonstrate human judgment. Anyone in the field can build these habits, computer science degree or not. The core move is the one Payton Grond made: treat AI as a genuine collaborator that expands what you can do, and keep its fluency separate from your own thinking.

The résumé and cover letter are arguments, not autobiographies. The word “résumé” comes from French, meaning to summarize, but by the 1940s, the meaning had morphed into the job document we are discussing here. These documents make specific claims: this person fits this position at this organization. Everything in them serves that claim or does not belong. This genre also stages the book’s central drama in miniature: AI can now produce a professional-sounding application instantly, which means that sounding professional is no longer enough on its own. What sets an application apart now is fit, specificity, and an honest picture of who you are. Learning Objectives Analyze a job posting as the rhetorical situation your materials must answer. Build a résumé around evidence-based accomplishment statements. Write cover letters that argue fit rather than restate the résumé. Use AI in the job search ethically and effectively, and represent yourself honestly. Title: Resume, Career, Jobsearch Author: resumaic Source: Pixabay License: Pixabay Content License Reading the Posting Treat the posting as your audience analysis. Highlight the required and preferred qualifications, the repeated terms, and the values language (“collaborative,” “fast-paced,” “mission-driven”). These are the criteria your reader, whether a person or an applicant-tracking system, will screen against. Your materials, if tailored well, can answer the posting point by point, in its own vocabulary (at least where it honestly applies), with your strongest evidence. The Résumé A résumé is a claims-and-evidence document, and it is usually skimmed in seconds, so design it to be read quickly: clean alignment, consistent formatting, informative section headings, and your most relevant material first. There is no single correct template, because conventions vary by field. This book teaches some of the current principles, but your campus career center is worth a visit for a field-specific review and more information about what has worked for recent graduates. Accomplishment Statements: XYZ and STAR The unit of persuasion on a résumé is the accomplishment statement, and a reliable way to build one is the XYZ method: accomplished X, as measured by Y, by doing Z. So “Reduced patient processing errors by 30 percent by redesigning the intake form” tells a lot more than “Responsible for forms.” The formula comes from Laszlo Bock, Google’s former head of People Operations, who described it in Work Rules! (2015), and it is advice Google’s recruiters still give applicants; for an accessible summary, see Bill Murphy Jr.’s account of Google’s résumé guidance for Inc. (external link). The point of the formula is to move you off duties and onto results: what changed, how much, and how you made it happen. The XYZ method has a close cousin you will meet again in interviews, the STAR method, which frames an accomplishment as Situation, Task, Action, and Result. STAR was built for interviews, where you have a minute or two to tell the whole story out loud, so it spends real time setting up the situation and the task. A résumé gives you no such room, and here the two methods come together: compress the story to a single line, imply the situation and task in a few words, and lead with the action and the measurable result. A strong résumé bullet is really a STAR story boiled down to its action and result, with a number attached, which is more or less what XYZ produces. Learn both, and use STAR to surface the raw material and XYZ to compress it for the page. Finding and Using Metrics Both XYZ and STAR put a number at the center of the accomplishment, the Y in XYZ and the result in STAR, and students often stall here, sure they have nothing to count because their work never produced a sales figure. Revenue is only one kind of measurement, though, and usually not the one most available to you. Most work produces metrics of some kind, if you know where to look. When you cannot point to dollars, you can still quantify along other dimensions. Scale and volume: how many documents you produced, customers you served, students you tutored, events you ran, or tickets you resolved. Frequency and consistency: how often, and how reliably, a weekly report or a semester without a missed deadline. Time: how long something took, or how much you saved, a turnaround cut from three days to one. Scope and responsibility: how big, a team of twelve you coordinated, or a budget or caseload you carried. Quality and outcome: what improved, an error rate you lowered or a satisfaction rating you raised. Reach and audience: readers, attendees, downloads. Selectivity and recognition: selected from a pool of applicants, placed in a competition, or chosen as one of a small number. Several of these measure scope or effort rather than results, and that is fine: “Coordinated schedules for a twelve-person team” quantifies the responsibility even though it doesn’t name a specific outcome. A before-and-after is often the most persuasive shape a student can reach for, because it shows change without demanding a dramatic figure, so it can be useful to name how things where and what resulted from your effort. Even when it seems difficult to get exact information, you can still make a good-faith effort to represent your work even when the numbers may not be perfect. When you don’t have an exact number, an approximation you can defend is fine. You can even write “approximately” and be ready to explain how you arrived at it. Count what you can actually count, roster sizes, event totals, shift hours, rather than inventing precision. And when a truthful number simply is not available, do not fabricate one; make the bullet concrete instead, naming what changed and who benefited, which is really the most impressive part, even without a number. The goal is not a number for its own sake but clear evidence a reader can trust, keeping in mind, of course, that numbers help the reader get a more memorable sense of scale and impact. Action Verbs Every accomplishment bullet should open with an action verb, a verb that names what you did: analyzed, designed, coordinated, built, led. Action verbs put you in the position of the person doing the work, and they push the bullet toward accomplishment rather than duty. “Responsible for scheduling” describes a job; “Coordinated schedules for a twelve-person team” describes you doing something. Match the tense to the time: past tense for past roles (“Analyzed,” “Organized”), present tense for a role you still hold (“Analyze,” “Organize”). For some reason, I often see students use some other tense like progressive (like Analyzing instead of analyzed. Not recommended). Avoid weak openers like “Helped,” “Assisted,” and “Worked on,” which quietly hand the credit to someone else, and try not to repeat the same verb down the page. The Purdue Online Writing Lab’s categorized list of action verbs for employment documents (external link) is a good place to look when you need a sharper word. Why Bullets, Not Paragraphs Résumés use bullet points, not paragraphs, and the reason is the reader. Recruiters skim, often in seconds, and a prose paragraph makes them hunt for the accomplishment buried inside it. Bullets isolate one accomplishment each, lead with the action verb, and line up down the page so the eye can scan them in parallel. They also parse more reliably for the applicant-tracking systems that read structure better than they read narrative. Save connected prose (full sentences and paragraphs) for the cover letter, where you tell the story; the résumé presents the evidence in scannable units. Relevant Coursework as Experience If you are still in your degree or newly finished, you may feel you have little to put on a résumé. You have more than you think. Coursework, and especially project-based coursework, is legitimate, relevant experience, and for an internship or an entry-level role in your field, it is often more relevant than the unrelated work experience you already have. A technical communication major who waited tables has real transferable skills worth listing, but a course project in which she designed and usability-tested a set of instructions speaks far more directly to a documentation internship than the restaurant job does. Recruiters hiring at the entry level know they are hiring for potential, and they read coursework as evidence of it. The key is to frame courses and course projects as accomplishments, with the same XYZ discipline you use everywhere else, rather than as a bare list of titles. You can place coursework in a few different spots, depending on how much weight you want it to carry. The lightest is a short “Relevant Coursework” line or list under your Education section, naming the courses whose content matches requirements in the posting, which is useful when a hiring manager is scanning for specific knowledge. The strongest is a set of bullet points under a named course or project, in an “Academic Projects” or “Relevant Experience” section, where you describe what you built, tested, or analyzed in accomplishment form: “Designed and usability-tested a twelve-step installation guide with five participants, cutting task-completion errors by 50%.” You can also fold the skills a course gave you into a Skills section, or narrate a course project as a short story in the cover letter, where prose can show fit. One rule to keep in mind: list a course only when it strengthens your case for this posting. Introductory and general-education courses rarely earn the space, while advanced, field-specific, project-heavy courses often do. For formatting patterns, see Indeed’s guide to including relevant coursework (external link). Ordering Education and Experience How you order your sections is itself a rhetorical choice, and the rule is to put your strongest, most relevant material where a reader will see it first. For a student or a recent graduate, that usually means education comes before experience, since your degree and your coursework are your best evidence. Once you have a position or two behind you, experience moves up and education drops down, listed in reverse chronological order (most recent first). Google’s recruiters give the same advice, and add a detail worth knowing: recent graduates should include school, degree, major, GPA, and graduation month and year, while the further you get from graduation, the less of that detail you need to carry. One caution the advice does not mention: a decades-old graduation year effectively announces your age, so at some point trimming that date may be a good idea. Skills, Templates, and Visual Résumés Résumé conventions vary by field, and a few format questions come up often enough to be worth addressing directly. If you work in a design-related field, a more visual résumé, one that shows off your sense of layout and type, is acceptable and sometimes expected, since the document itself works as a sample of what you can do. In most other fields, a conventional layout will probably work better than trying to look fancy. Be careful with templates. A template can help you get started, but many are built to fill space rather than to make an argument, and they nudge you toward sections you don’t really need. This can work against you. A decorative sidebar or multi-column grid can confuse the applicant-tracking systems that parse a single clean column more reliably, and templates tempt you to pad, encouraging you to fill up the space in, say, a skills sidebar. That brings up the skills section, which is worth thinking through rather than copying from a template. A dedicated skills section works best as a scannable home for concrete, verifiable skills: the specific tools, software, languages, and certifications a recruiter or an ATS can match against the posting. These are often what people have traditionally called “hard skills,” though many are also the perishable skills discussed in the sense of Chapter 1, tied to particular tools and technologies that can go out of date. Durable skills, the human-centered abilities like collaboration and problem-solving, do little work sitting in a list, and generic entries like “team player” or “strong communicator” read as filler rather than evidence. The better move is to prove a durable skill in a bullet, with something that actually happened, rather than to assert it. Employers say they want these durable skills (see the NACE Career Readiness Competencies (external link), also discussed in Chapter 1), but claiming one does little on its own; showing it through a real accomplishment is what carries weight with a reader, which makes a pretty good argument for putting them in the context of a bullet point where you can “show” rather than “tell.” The Objective and the Summary Statement Many résumé examples open with a summary statement. Traditionally that space held an objective statement, which announced the kind of job the candidate wanted. Most career advisors now treat the objective as outdated, because it spends the résumé’s most valuable space on what the candidate wants rather than on what the employer needs; Google’s recruiters simply tell applicants to skip it, on the reasoning that they already know your objective is to land this job. The summary statement replaces it with a brief, achievement-focused pitch, three or four lines or a short cluster of bullets, that leads with your strongest, most relevant, ideally quantified accomplishments and signals your fit for this specific role. A summary works best, though, when you have a record to summarize. With five or more years of relevant experience, it earns a space on your document. For a soon-to-be or recent graduate it often does not, because with little to condense, the summary tends to fill with generic phrasing: “Passionate, hard-working professional looking to leverage skills.” That can read as filler to a recruiter giving your résumé a six-second scan. Space is scarce, and padding it, by inflating detail, enlarging fonts, or stretching white space onto an awkward second page, works against you. For most new graduates the better move is to skip the summary, keep the résumé to a tight single page, and let strong, tailored accomplishment bullets carry the argument. Include a summary only when it can act as a data-backed elevator pitch that proves your fit in a few seconds. If you catch yourself writing generic phrases in your summary, you might just want to cut it. The Cover Letter The letter argues fit in three moves. Open by naming the position and your central qualification for it, specifically, not just “I am writing to apply.” In the body, develop two or three claims that connect your evidence to the posting’s priorities, adding what the résumé cannot: context, motivation, and knowledge of the organization, since one genuine, specific detail about the employer outperforms a paragraph of enthusiasm. Close by restating fit and pointing forward. Keep it to one page, and match the register of the industry you are writing to. Some industries want very short cover letters; others may want longer ones; others may not expect them at all. It’s probably better to have no cover letter at all than to have a poorly-written one. Co-Intelligence Workflow: Employment Materials I mentioned that AI can produce these job documents. LLMs were trained on large amounts of data from the internet, including sample resumes and tons of career advice websites. But giving the process over too fully can make you miss out on the reasoning process that condensing to bullet points requires of you. With that in mind, here are some ideas for how to approach AI. You supply: the real posting, your honest inventory of experience and skills, and the facts about the employer you actually gathered. AI drafts: a first-pass tailoring of your résumé bullets to the posting’s language, and a rough cover letter from your materials You diagnose: mark every sentence that could appear in anyone’s letter because it’s generic. That generic register could apply equally to your competition’s letter too. Replace it with specifics only you can claim. AI assists revision: tighten, check parallelism, and simulate a skeptical recruiter (“what would make this reader doubt fit?”). You verify and own: every claim true, every detail accurate, and disclosure handled according to your context’s rules. You will sit in the interview alone, so the materials have to be yours. Durable Skills in This Genre Self-knowledge and narrative judgment: AI can phrase your experience, but it cannot know which experiences matter or what story they tell together. Integrity: the discipline to stay accurate under pressure to embellish. Fortitude: applications are a rejection-rich environment, and persisting through it, revising rather than despairing, is itself a professional skill. Practice it now, in the low-stakes version. Reflect: Peer + AI Review Exchange drafts with a peer; separately, ask an AI reviewer to evaluate your materials against the posting. Then write one page: Where did the two reviews agree and disagree? Which peer comments could no AI have made, and which AI comments did your peer miss? What did you change, what did you decline to change, and why?