TSMC Co-COO Compares AI to Powerful Yet Immature Three Year Old Superman

Taiwan Semiconductor Manufacturing Company, the world’s largest contract microchip maker, has urged technology leaders to adopt a cautious approach when applying AI to complex semiconductor research.

Speaking to engineers and computer science students, the company’s executive leadership explained that while smart software tools excel at processing vast datasets and automating standard computer code, they remain inherently immature, unpredictable, and potentially dangerous when handling secret technological formulas.

Because microchip manufacturing relies on highly confidential scientific research, using unmonitored artificial intelligence systems risks leaking sensitive technical secrets or making costly engineering errors.

The executive warning was delivered by TSMC Co-Chief Operating Officer Y.J. Mii during an academic forum at National Taiwan University in Taipei on Tuesday, September 15, 2026.

Addressing questions about how the global chipmaking giant integrates automated tools into its operations, the senior executive detailed where software automation succeeds and where human expertise remains mandatory.

While software helpers significantly speed up routine circuit design and basic programming tasks, TSMC strictly limits how much automated tools participate in advanced research and manufacturing decisions.

The main reason TSMC is urging caution with AI right now is that advanced computer chips have reached physical limits where standard formulas no longer exist.

When developing next generation microchips smaller than seven nanometers, engineers encounter uncharted scientific territory with unknown physical variables.

Because AI models rely heavily on existing historical data, they cannot solve entirely new physical problems and lack the ethical judgment required to distinguish between correct decisions and dangerous mistakes.

Furthermore, because commercial software programs process user inputs through external servers, using them for high level research exposes priceless manufacturing blueprints to potential corporate espionage or security leaks.

Describing the current state of artificial intelligence software using a vivid real world comparison, TSMC Co-Chief Operating Officer Y.J. Mii stated during the university forum that “AI is like a 3 year old Superman, powerful but immature, unable to distinguish right from wrong, and potentially capable of causing damage”.

Explaining why the semiconductor giant enforces strict safety boundaries around proprietary research files, Y.J. Mii noted that TSMC takes a cautious approach to automated tools primarily because “AI may leak sensitive information,” which represents a major security risk for a company whose global edge rests entirely on advanced research.

Emphasizing that major engineering choices must always rely on human collaboration rather than automated software outputs, Y.J. Mii added that “AI can compute data but cannot make decisions,” stressing that final manufacturing choices must always come from human discussions among expert teams.

By framing artificial intelligence as a powerful yet immature tool, TSMC is reminding the global tech industry that human wisdom remains indispensable.

Maintaining strict safety controls over software automation ensures that microchip innovation advances safely without endangering critical technological secrets.