CairnCareers vs asking ChatGPT.
Most students try the chatbot first. It is free, it is fast, and it will happily tell you which jobs are safe from AI. The problem is that it cannot show you where any of that came from. This page sets out what each one gives you, and when the chatbot is the right tool.
| Feature | ChatGPT | CairnCareers |
|---|---|---|
| Where the answer comes from | Patterns in training data. The same question can get a different answer tomorrow, and there is no source to check. | Each career path is scored against O*NET, the U.S. Department of Labor's occupational task data, using the Eloundou et al. (2024) exposure framework published in Science and METR's* capability trend.1 |
| Can you check the numbers | No. A chatbot can name a study, but it cannot show that the number it gave you came from it. | Yes. Every source is named and linked on the methodology page, and every number on the dashboard carries a footnote.1 |
| Pay and growth data | Mixed into the same answer as everything else, from whatever the model remembers. | From the U.S. Bureau of Labor Statistics, shown as context, and kept structurally separate from the exposure score so you can see what moved the result.1 |
| What happens when you have no number | It will often write one for you, which is how invented figures end up on resumes. | The coach asks the one question that would produce the number, and if you do not have one it falls back to an honest description of scope. It never invents a figure. |
| The plan | A list of suggestions in one reply, with no memory of what you did last term. | A term-by-term roadmap tied to the academic calendar, each term closing one evidence gap, plus resume, LinkedIn, network, and interview modules that share the same records.3 |
| Built for | Everyone, for everything. | College students and recent graduates choosing a first path. |
| Price | Free tier, with paid plans set by OpenAI. | Free tier. Pro at US$6 a month. Premium at US$11 a month.2 |
When ChatGPT is the better tool
- Drafting a cover letter, rewording a bullet you already have a real number for, or rehearsing answers to interview questions out loud.
- Quick, low-stakes questions where you do not need to defend the answer to anyone.
- You can use both. Bring the sourced numbers from your dashboard to the chatbot and let it help you with the wording.
When CairnCareers is the better tool
- You are choosing between paths and want to know which tasks in each one are exposed to AI, with a source you can hand to an adviser, a parent, or a recruiter.
- You need a plan that survives past one chat window and moves the number over a year.
- You want the number on your resume to be one you can back up.
Is ChatGPT good for career advice?
For wording and practice, yes. For deciding which career path holds up to AI, it has a gap: it cannot show where its answer came from, and the same question can get a different answer tomorrow.
That gap matters most when the answer goes somewhere. If you repeat a chatbot's claim about a job to an adviser or a recruiter, you have nothing to point to when they ask for the source. A score with a footnote gives you something to point to.1
How is the AI-exposure score calculated?
It averages how exposed each task in an occupation is to current AI, weighted by how much of the work each task takes up, and puts the result on a 0 to 100 scale.1 The tasks come from O*NET. Each one gets an exposure value from the Eloundou et al. (2024) framework: 0 if today's AI cannot meaningfully do it, 0.5 if AI can do it with the right tools, and 1.0 if AI can do it on its own.1
A worked example: half of an occupation's work is a task AI can do on its own, a quarter is a task AI can do with tools, and a quarter is a task AI cannot touch. That gives (0.5 x 1.0) + (0.25 x 0.5) + (0.25 x 0) = 0.625, or about 63 out of 100.1
What happens when you do not have a number for a resume bullet?
The coach asks for the number instead of writing one. Take the sample from the home page: a marketing class group project that ran a campus survey for a local coffee shop.
Ask a chatbot to add metrics and it may supply a response count whether or not anyone counted. The coach asks how many people answered. If you do not know, the bullet describes the scope honestly instead: designed and ran a customer survey for a local business and presented pricing recommendations to the owner.
How do you use ChatGPT and CairnCareers together?
Use each one for the part it does well. Start on the dashboard, where the AI-exposure score, the task map, and the pay figures each carry a footnote you can open.1 Then take a finished resume bullet to the chatbot and ask it for tighter wording, or ask it to play the interviewer while you practice the answer out loud.
Keep one rule when you do: the chatbot may change the words, never the facts. If the survey bullet comes back with a response count or a percentage that was not in your record, delete it, or answer the coach's question and put the real figure in yourself. A number you cannot explain in an interview costs you more than a plain bullet does.
Why are pay and growth kept out of the AI-exposure score?
Because a score that mixes them stops telling you about AI. Pay and growth come from the U.S. Bureau of Labor Statistics, the Occupational Outlook Handbook for growth and Occupational Employment and Wage Statistics for pay, and they sit beside the exposure score without moving it by a single point.1
A chatbot answer tends to fold all three into one verdict, such as a job that is safe, well paid, and growing. When the verdict changes, you cannot tell which part moved. Kept apart, a well-paid path can still show high exposure, and you weigh the two yourself instead of inheriting someone else's trade-off.1
What do you need to give CairnCareers to get a result?
No resume and no work history. You bring the career paths you are weighing and the work you have already done: classes, group projects, jobs, and clubs.1 Each path is matched to a standardized occupation, and you see the plain occupation name and how close the match is.1
The Free tier returns the AI-exposure score, a map of which tasks are durable and which are exposed, and median salary context.2 Pro rewrites your classes, projects, and jobs as resume bullets and re-runs your scores each month. Premium adds a resume and LinkedIn that update as you add experience, who to talk to next and what to ask them, and a term-by-term roadmap to graduation.2 A chatbot asks for nothing up front either, but it also keeps nothing, so next week you start again from a blank window.
Why not just ask ChatGPT to cite its sources?
You can, and it will usually give you some. The trouble is checking them. A chatbot can name a study that does not contain the figure it gave you, or a report that does not exist, and the only way to tell is to open each one and look for the number yourself.
The dashboard works the other way around. The score is computed from the sources, not written first and sourced afterwards: O*NET supplies the tasks, Eloundou et al. (2024) supplies each task's exposure value, and the arithmetic is shown on the methodology page with a worked example.1 Each footnote opens the source the number came from.
Sources
- CairnCareers methodology page: the three sources behind the score (O*NET, Eloundou et al. 2024 in Science, METR) and the 0 to 100 scale.*
- CairnCareers pricing: Free, Pro at US$6 a month, Premium at US$11 a month, on the pricing section of the homepage.
- CairnCareers sample roadmap: the term-by-term plan, one evidence gap per term.
Statements about ChatGPT describe its public pages on the date checked and may have changed since, so if something here is out of date, tell us and we will correct it.
* This projection uses METR's established long-run time-horizon trend, not the faster 2026 estimates. Those newer figures come from a task suite near saturation, carry very wide confidence intervals, and measure software tasks specifically, so extrapolating them across all occupations is not yet reliable. See MIT Technology Review, "This is the most misunderstood graph in AI," February 5, 2026, https://www.technologyreview.com/2026/02/05/1132254/this-is-the-most-misunderstood-graph-in-ai/