In Plain Sight - 10
In Plain Sight is a weekly blog from Rhys that we will publish every Thursday/Friday. Each edition will be focused on a thorny topic within investing, startups, learning and work and policy.
In Plain Sight is partially a personal attempt to think independently and to avoid leaning too much on AI for answers to our questions. Will we use AI to edit and polish the text? Yes. But more importantly, will we come up with all of the ideas and analysis? Yes.
Here follows Edition 10 - 'Every qualification is going to be rewritten'
------
In a recent piece, we asked whether doctors still need a decade to qualify. Our answer was that the better question is not whether to train doctors for less time, but whether to train them differently.
We'd extend that argument well beyond medicine. Most professional qualifications were designed around three assumptions: that the job stays broadly the same, that the knowledge you need can be memorised up front and that it is best delivered in a classroom before you start work.
All three are now moving at once. The job is changing as AI takes on routine tasks. The knowledge that matters is shifting from recall to judgement. And knowledge itself is increasingly delivered at the point of need, inside the tools people use every day.
So which qualifications break first? We think the most exposed share two traits. Their exams mostly test knowledge that is written down. And the junior work that used to train people is the work AI is now absorbing.
The key point for us is simple: a qualification can certify what you know at a static point in time, or what you can do with what the machine knows on an ongoing basis. With this in mind, below you'll find six mini case studies, each with our hypothesis for how the qualification evolves.
1. Law: a new form of traineeship?
In England and Wales, the route to qualifying has already been rebuilt once. Since 2021, the Solicitors Qualifying Examination has separated qualifying from a law degree, paired with two years of qualifying work experience that can be split across up to four organisations.
The pressure now sits in that work experience. Trainees have traditionally learned through document review, due diligence, disclosure and first drafts. That is exactly the work legal AI tools now do in minutes.
Our hypothesis: the exam persists, but the two years change shape. We expect supervised, AI-assisted casework and simulated matters to count towards experience, with assessment focused on judgement: spotting the issue the model missed, advising a nervous client and knowing when a draft is wrong.
2. Accountancy: from preparing the numbers to judging them?
Accountancy is the clearest example of a profession already adjusting (loosening?) its entry rules. Facing a talent shortage, more than 40 US states have now adopted a route to CPA licensure that swaps the 150-credit-hour rule for a bachelor's degree plus two years' experience.
The exams still lean heavily on technical recall and reconciliations, audit sampling and tax preparation are among the most automatable tasks in professional services.
Our hypothesis: the next reform is to what gets tested. We expect less weight on rules and more on professional scepticism, data interrogation and client advisory, with assessments that hand candidates an AI-prepared set of accounts and ask them to find what's wrong.
3. Teaching: will the deliverer of knowledge be redefined?
Teaching is the case closest to home for us. If AI tutors can explain a concept, set practice and mark it, then more of a pupil's contact with content happens outside the teacher's direct delivery.
That doesn't make the teacher less important. It changes what the job is: diagnosing why a pupil is stuck, sustaining motivation, managing a room, and deciding when to lean on the tool and when to put it away. Most teacher training still prepares people mainly to plan and deliver lessons.
Our hypothesis: teacher qualifications add orchestration as a core competency. We expect training to include working alongside AI tutors, reading learning data and designing lessons where the machine handles practice and the teacher handles the thinking.
4. Software engineering: the degree takes a new shape?
Software has no licensing body, which makes it an early warning for everyone else. The market adjusts before any regulator or assessor changes what needs to be known.
And it is adjusting. Stanford's Digital Economy Lab finds that employment for 22 to 25 year olds in AI-exposed occupations, including software development, is now 19% below where it would have been had it tracked less-exposed peers, mostly through reduced hiring. The researchers are careful to call this descriptive rather than causal. But when AI writes a growing share of first-draft code, the junior role that turned graduates into engineers is thinner.
Our hypothesis: the computer science degree evolves. We expect more weight on architecture, code review, security and specifying problems well and employers leaning on portfolio and work-sample assessments over the degree itself.
5. Investment and financial advice: a new shape for accountability?
The CFA Institute in the US is a useful example of a professional body already making updates. It has added practical skills modules in areas like Python and financial modelling and for exams from February 2027 it is introducing a module on financial data science, AI and large language models, plus new content on communicating investment insights.
This feels like the right direction, but the core of the qualification is still a knowledge test that a model can increasingly pass.
Our hypothesis: the valuable part of the credential becomes ethics, communication and accountability. When a model can build the valuation, what clients and regulators need is the person who will stand behind it, explain it plainly and carry the fiduciary duty...
6. Pharmacy: a new focus on decision-making?
Pharmacy shows what a rewritten qualification looks like in practice. From summer 2026, most pharmacists joining the UK register are independent prescribers from day one, with prescribing built into the degree and foundation year rather than added years later.
The job moved first in this case. Dispensing is increasingly automated and centralised, while pharmacists take on more clinical consultation. The qualification then followed.
Our hypothesis: pharmacy becomes the template. We expect the next step to be continuous, workflow-embedded assessment, with decision-support tools doing the checking and training focused on consultation, prescribing judgement and knowing when to refer.
The pattern
Across all six, the same shift repeats relatively consistently. A qualification stops being a snapshot of knowledge taken once and becomes something closer to a licence to exercise judgement that is renewed and evidenced over time.
The risks are significant. If junior work disappears faster than simulation replaces it, we risk a generation that can supervise AI output without ever having built the underlying intuition. Professional bodies move slowly and qualifications that change too fast can lose the public trust that gives them value. And assessing judgement is harder, and more expensive, than assessing recall.
That is where we see the opportunity: simulation that replaces lost apprenticeship hours, assessment that measures judgement credibly and learning embedded in the tools professionals already use.
If you're building in this space, we'd love to talk to you.
.png)
