Discovery’s CIO: AI coding gains are real but far from 100%
The boldest AI coding predictions say developers will be automated out of their jobs within a few years. Derek Wilcocks, group CIO at Discovery, is not buying it yet, and he has the internal measurement to explain why.
What Discovery is actually seeing
Speaking on TechCentral’s Meet the CIO podcast, Wilcocks said Discovery is measuring efficiency gains of roughly 20-25% in some areas from AI coding assistants. But he was quick to add context: developers in large corporates typically spend only 25-30% of their time writing code. The rest goes to meetings, code reviews, testing, and sessions with business users. A 25% gain on 25% of the job is not the headline number vendors tend to advertise.
Discovery builds most of its own software in its biggest businesses, including health and life, and buys little traditional enterprise software. That makes the quality of AI-generated code a real operational concern, not an abstract one.
Three unsolved problems
Wilcocks named three blockers keeping fully agentic AI coding out of large enterprises right now: explainability, efficiency, and maintainability of AI-generated code. None is solved today, he said. Anyone who has run a significant project on AI-generated code knows that a year later, after model drift, they are left with something even an AI is not sure the previous AI wrote. That is a serious risk in an environment where code is the systemisation of a business’s core processes, not just a set of instructions.
The harness, not the model
Discovery’s senior developers use all the major agentic coding frameworks, but the group has invested heavily in constraining what those tools can do. The work has gone into limiting agent actions so output meets Discovery’s coding standards, security requirements, privacy rules, and compatibility with existing systems.
Wilcocks’ broader prediction is that the companies that win on AI will not be the ones with access to the best frontier model. They will be the ones that build the harness: the scaffolding that lets AI be deployed safely and at scale. That is where he thinks the real competitive advantage gets built.
Hiring and retraining
The ability to write code alone is no longer enough to walk into a developer job, Wilcocks said. What still matters alongside coding skill is the grounding a computer science degree provides: how architectures work, how to think about algorithms, design-led approaches, risk, and governance. Communication and the willingness to keep learning round out the profile he is hiring for.
On retraining, the numbers surprised him. Discovery made online learning platforms from well-known providers freely available to its IT staff. He had expected 3-4% uptake. Instead, 50-60% of staff completed at least one AWS or Google programme. Hiring a specialist in agentic security or advanced AI models is close to impossible right now, he said. Training graduates who can learn quickly is the practical alternative.
For startups, the calculus is different. Wilcocks said a company founded today would have its head in the sand if it were not generating its code almost entirely with AI. The caution applies specifically to large regulated enterprises touching multiple legacy systems across multiple teams. He noted this is a pattern he is hearing from other CIOs as well.
I’m founder of BizStack at Cagri Sarigoz LLC and a passionate advocate for entrepreneurs.
With over 14 years in tech, marketing, and AI, including my role as Head of SEO at CitizenShipper and co-founder of TaleBot at Intale AI, I’m dedicated to sharing genuine, useful product insights and tips.
At BizStack, I aim to cut through the digital noise to provide clear, actionable advice.
And more than all else, I’m a father to a (always) little girl and a husband.
Contact me at [email protected] for assistance.
