🔍 Read the full analysis: What AI Means For Learning To Code: Evidence For Students And Professionals on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Evidence available in 2026 does not show that learning to code is pointless, but it does show that AI is changing which coding skills matter and may be narrowing entry-level opportunities. AI can help with some tasks, yet productivity gains remain uncertain and one learning experiment found lower quiz scores among developers who relied on AI.
AI is writing a growing share of software code, but evidence available in 2026 does not show that learning to program has become pointless. Company disclosures, controlled tests and hiring data instead point to a changing role for developers: directing and reviewing generated code matters more, while the effects on productivity, learning and entry-level jobs remain unsettled.
Anthropic said that more than 80% of the code merged into its codebase in May 2026 was authored by Claude, compared with what it described as a low-single-digit share before its Claude Code tool launched in February 2025. The company also reported that its typical engineer merged eight times as much code per day in the second quarter of 2026 as in 2024. Anthropic cautioned that lines of code measure quantity rather than quality and that the eightfold figure is “almost certainly an overstatement” of the productivity gain.
Independent evidence on speed is less conclusive. In a randomized METR trial involving 16 experienced open-source developers and 246 tasks, participants took 19% longer when allowed to use AI tools available from February to June 2025. They had expected AI to make them 24% faster and, after the work, believed it had made them 20% faster. METR said a later study offered only “very weak evidence” about the size of any improvement by early 2026.
Hiring figures raise a separate concern for people starting out. Indeed data cited in the source show that 4.5% of U.S. software-development job postings were entry-level in the first quarter of 2026, while 69.3% were senior. Stanford researchers found employment among 22- to 25-year-olds in AI-exposed jobs 19% below where it would have been otherwise. Yet U.S. software-development postings on Indeed were up about 23% year over year in November 2026, after reaching a low in May 2025. These measures describe different things and do not, by themselves, establish that AI caused the employment changes.
The Skills Shifting in Software Jobs
For students and career changers, the evidence supports learning programming fundamentals while adding skills in specifying tasks, reading code, testing and debugging. If AI produces more of the initial code, workers still need to judge whether it fits the system, works as intended and creates security problems. That judgment depends on understanding code rather than treating a generated answer as reliable by default.
Employers face a related question: whether to hire and train early-career developers when AI tools can handle some implementation work. The low share of entry-level postings is a warning about access to first jobs, but the available figures do not prove that automation is the cause. For professionals already in the field, measured productivity will depend on task type, experience, codebase complexity and how output quality is counted—not just the volume of code produced.
Products Worth Considering

Logitech MX Brio Ultra HD 4K Webcam

Kindle Paperwhite Signature Edition 32GB (2024 Model) – Metallic Black

DJI Mini 5 Pro Fly More Combo with DJI RC 2, 1-Inch CMOS
As an affiliate, we earn on qualifying purchases.
From Code Generation to Review
In March 2025, Anthropic chief executive Dario Amodei predicted that AI would write 90% of code within three to six months and essentially all of it within a year. The company’s later internal figures suggest substantial use at Anthropic, though the 90% prediction was not established as a measure of the wider software industry. Amodei also said programmers would still need to specify what was being built, how it fit with other code and whether its design was secure.
Other figures provide limited points of comparison. Google chief executive Sundar Pichai said in October 2024 that AI generated more than a quarter of new code at Google, which engineers then reviewed and accepted. METR cited an estimate in February 2026 that Claude Code authored about 4% of GitHub commits. These figures cover different organizations, periods and measures, so they should not be treated as a single industry-wide estimate.
Research on speed also varies by task. A 2022 experiment with 95 professional developers found that those using GitHub Copilot completed a small JavaScript HTTP-server task 55.8% faster on average. That result differs from METR’s later study of real tasks inside large, familiar codebases. The experiments therefore do not establish one general productivity effect for all programming work.
““More than 80% of the code we merge into Anthropic’s codebase was authored by Claude.””
— Anthropic
What the Evidence Cannot Yet Show
There is no settled estimate of how much AI increases software productivity across the industry. Anthropic’s internal figures come from a company that develops and uses its own AI coding tools; its code-volume measure is not a direct measure of output quality. METR’s first controlled trial involved a small group of experienced developers working in codebases they already knew, while its later study had confidence intervals that included no effect.
The learning evidence is also limited. In an Anthropic experiment, mostly junior developers using AI while learning a new library scored 50% on a follow-up quiz, compared with 67% for developers who coded by hand; the largest gap was in debugging. The supplied account does not establish whether this result generalizes to other tools, teaching methods or longer-term learning. Nor do the job-posting and employment figures establish that AI caused changes in early-career hiring.
Evidence to Watch in Hiring and Training
The next useful evidence will need to measure more than how much code an AI tool generates. Longer-running studies could track task completion time, bugs, code quality and retention of skills, across different experience levels and types of software work. Better comparisons of hiring and employment trends could also help distinguish AI effects from broader changes in the labor market.
For now, students and professionals can treat AI as a tool to learn alongside rather than a replacement for understanding code. The studies and company disclosures reviewed here point to continued demand for people who can define problems, inspect generated work and find defects, while leaving open how quickly those skills will affect hiring and productivity.
Key Questions
Is learning to code still worthwhile?
The evidence does not show that it is a waste of time. It suggests that programming knowledge remains useful for specifying tasks, reviewing AI-generated code, testing behavior and identifying bugs or security weaknesses.
Does AI make professional developers faster?
Results depend on the task. METR’s 2025 trial found experienced developers took 19% longer with AI on the tested work, while a 2022 Copilot experiment found faster completion of a small coding task. METR said its later results were too weak to establish the size of any subsequent speed-up.
What did the Anthropic learning experiment find?
Mostly junior developers who used AI while learning a new library scored 50% on a follow-up quiz, compared with 67% among those who coded by hand. The experiment’s results do not establish that all AI-assisted learning leads to lower scores.
Are entry-level software jobs disappearing?
The cited Indeed figures show entry-level roles accounted for 4.5% of U.S. software-development postings in the first quarter of 2026, while senior roles accounted for 69.3%. The figures do not prove that AI caused the pattern, and postings were up about 23% year over year in November 2026.
Source: ThorstenMeyerAI.com
