Skip to content
QIO journal

AI and automation4 min read

Computer science graduates: the share of coders fell from 40 to 28 % in a year

The UK Graduate Outcomes Survey, which tracks graduates' careers 15 months after university, recorded a sharp drop: the share of computer science graduates starting work as programmers fell from 40 to 28 % in a year. Here are the numbers, the researchers' explanations and the role of generative AI.

Illustration for the study on computer science graduates

The numbers

Three indicators from one dataset.

  • Share of CS graduates starting as coders or programmers: from 40 to 28 % in a year.
  • Share of graduates in professional-level jobs: from over 60 to 50 % in two years.
  • Junior developer hiring per LinkedIn and UK government data: down 27 % year on year.
A staircase missing its bottom step and a small figure at the foot
The bottom step disappears: tools take over typical junior tasks.

We see the same picture in website work: standard forms, integrations and tests take a model minutes, and the specialist's value moves to framing the task and checking the result. That is exactly what beginners should learn.

What the researchers say

The authors are careful about causality. They name several factors: a general cooling of the labour market, corporate caution after pandemic over-hiring, expensive capital. Generative AI may have amplified these processes, but the data does not prove a direct causal link. That matters: headlines such as 'AI killed juniors' run ahead of the evidence.

Where AI fits

The mechanism the researchers describe works like this: generative models cover the typical tasks of beginner developers, boilerplate code, tests, simple interfaces. Those tasks used to be the training ground for newcomers. Companies hire fewer juniors not because models replaced engineers but because the work on which engineers grew has disappeared.

What it means for the market and education

For graduates: value shifts from writing code to framing tasks, checking results and integrating systems. For companies: if the bottom step is gone, build your own with mentored internships on real tasks, or in five years there will be nobody to hire at middle level. For education: teach working with models as a tool rather than banning them.

Summary

The UK data records a drop in entry into the profession, not the death of the profession. AI is one factor, not a proven cause. The market is rebuilding its bottom step, and those who build it themselves win.

Sources

  1. Fewer programmers, starting with beginners (SecurityLab, in Russian)
  2. HESA: Graduate Outcomes
  • Graduate Outcomes Survey
  • Generative AI
  • Junior developer
  • LinkedIn

Follow the journal

New pieces on websites, SEO and AI come out in the QIO journal. Follow in Google, by RSS or in Telegram to get them first.

Read next

How can we help?
Discuss a project