Pragmatic by design: Engineering AI for the real world
A new MIT Tech Review report reveals how product engineers are cautiously scaling AI to avoid real-world risks.
A new report from L&T Technology Services and MIT Technology Review Insights reveals a pragmatic, risk-averse approach to AI adoption within product engineering. Surveying 300 senior technology executives, the research found that while 90% of engineering leaders plan to increase AI investment over the next 1-2 years, the growth is measured. The largest group (45%) plans increases of only up to 25%, prioritizing gradual trust-building over radical transformation. This caution stems from the high-stakes nature of physical products, where AI-informed design errors can lead to safety recalls or structural failures with irreversible real-world consequences.
Consequently, engineers are deploying layered AI systems with distinct trust thresholds instead of general-purpose models. The top investment priorities are predictive analytics and AI-powered simulation/validation, chosen by a majority for their clear feedback loops and ability to prove ROI and secure regulatory approval. The report emphasizes that verification, governance, and explicit human accountability are non-negotiable. Measurable outcomes are focused on customer-visible signals like product quality and sustainability (e.g., defect rates, emissions) rather than internal operational gains or pure innovation speed.
- 90% of product engineering leaders plan to increase AI investment, but 45% will boost budgets by only up to 25%, showing cautious, incremental adoption.
- Predictive analytics and AI-powered simulation are the top investment areas, valued for providing auditable performance data and clear ROI proof points.
- Engineers mandate human oversight and use layered AI systems with trust thresholds to manage physical risks, prioritizing product quality and sustainability over speed.
Why It Matters
This shift means safer, higher-quality physical products—from cars to medical devices—as AI is integrated with rigorous human oversight and measurable outcomes.