People keep asking me this question.
Sometimes directly. Sometimes through polite questions of a conference panel. And recently, through the renewed debate around Sam Altman and claims that he does not code.
Whether those claims are fair or simply the latest proxy war in tech commentary, I do not want to wade into the specifics. What I do want to engage with is the real question underneath: what kind of leadership actually builds trustworthy AI?
Because I have a view. A lived one.
Let me be clear about what coding literacy means at CEO level. It does not mean writing production code at midnight. It means governing risk intelligently.
Traditional software failures are inconvenient. AI failures can be systemic. Hallucinations. Data leakage. Embedded bias. Poor explainability. Governance gaps that look fine on paper and catastrophic in practice. In regulated sectors, these are not edge cases. They are foreseeable.
When you understand how models behave under pressure, what training data provenance actually means, and where safety guardrails are substantive versus performative, you ask different questions. Better questions. Earlier.
Technical fluency also compresses the distance between vision and execution. I can speak directly with my engineering and ML teams, challenge over-simplification, and align roadmap decisions to technical reality without micromanaging the brilliant people doing the work. That matters enormously.
And there is something else. Technical literacy acts as a brake on AI-washing. When a leader cannot interrogate a claim like 'the model is accurate' or 'the model is compliant', organisations fill that gap with vague assurances, outsourced accountability, and impressive demos backed by weak audit trails. I have seen it. It worries me deeply.
Engineering cultures respond differently when leadership understands constraints. Customers and regulators respond differently too. Technical fluency builds trust faster in high-stakes environments, not because it signals intelligence, but because it signals genuine competence.
My ability to code, and my practical understanding of machine learning, has directly shaped the innovation culture within GAI Labs.
Not because I impose technical decisions. But because the design conversation at Guildhawk starts from a different place. It starts with security, with human accountability and auditability. With the multilingual reality that most AI systems still handle poorly.
That thinking is why we built a one-click human review option alongside our AI translation. Because I understand exactly where automation adds value and where human certification is not optional. It is essential.
That distinction is everything.
I believe in intellectual honesty, so let me give the counter-argument its full weight.
A CEO's core responsibilities are strategy, capital allocation, culture, partnerships, and governance. Coding is not the same as leadership. A CEO can be an exceptional operator without writing a single line of code.
There is also a real risk that 'technical CEO' becomes a vanity metric, driven more by internet culture wars than boardroom evidence. Knowing how to code does not guarantee sound risk management, ethical governance, or product-market fit. I have watched technically brilliant people make catastrophic strategic decisions.
Partial fluency can create false confidence too. A CEO who knows 'enough to be dangerous' may overrule specialists, simplify problems prematurely, or back architectural decisions based on personal preference rather than evidence. Occasionally, a non-technical leader asking 'naive' questions exposes deeper flaws precisely because they are not seduced by technical detail.
And AI leadership is now as much political, legal, and societal as it is technical. AI touches workforce disruption, public legitimacy, national security, regulatory liability, and global competition. A CEO fluent in diplomacy, regulation, and public accountability may outperform a technically brilliant leader who lacks those capacities.
I hold both of these things to be true at the same time.
Not every AI CEO needs to be a daily coder.
But AI CEOs do need sufficient machine learning literacy to govern risk, challenge assumptions, and prevent the kind of AI-washing that is quietly becoming one of the industry's most serious problems.
In regulated, multilingual, high-trust environments where outcomes affect real people and compliance carries real consequences, the right questions are not technical in the narrow sense. They are governance questions.
Answering those questions well requires understanding the systems you are asking society to rely on. At Guildhawk, that understanding has helped us build AI that is secure by design, transparent where it matters, and paired with human expertise when the stakes demand it.
The real test
The debate about whether a CEO can code is ultimately a proxy for something deeper.
Technical literacy matters less as a badge and more as a capability. The measure is not whether a leader can implement a neural network. The measure is whether they can govern what they build, with rigour, with accountability, and with the wisdom to know where human judgement remains irreplaceable.
In an era where AI is becoming infrastructure, that capability belongs at the top.
Jurga Žilinskienė MBE is CEO of Guildhawk and GAI Translate, and founder of GAI Labs.