![Adam Mackay](assets/images/adam_mackay.jpg) # Adam Mackay *Writer, researcher, and strategist on AI, autonomy, and the physical world.* I explore what happens when AI moves from advising people to acting in the world, and how organisations decide when these systems are trustworthy enough to use. My work draws on 28 years across safety-critical industries. [Discuss a research or writing project](/contact) [Read the Newsletter](/newsletter) ## What I Do Currently I'm **[Head of AI and Principal Strategic Advisor at QA-Systems](https://linkedin.com/in/adammackay)**, looking at how language models change the economics and safety calculus of testing safety-critical software. The research runs wider than the role. Aerospace, automotive, medical devices, and defence are the sectors where release pressure meets certification evidence, and where the cost of being wrong is measured in something other than revenue. That is the ground my writing works over. Co-author, *[Embedded Software Testing](/book)* (BPB Publications, 2026) · Cover feature, *[The Measurement Gap](https://www.heise.de/select/ix/2026/6/2609809170334426604)*, [heise iX](https://www.heise.de/select/ix/2026/6) 06/2026 · Contributing author, [HiPEAC Vision 2025](https://www.hipeac.net/vision/2025/) and [2026](https://www.hipeac.net/vision/2026/#/) --- ## Current Research: Safe Enough ![Safe Enough research desk with working notes](assets/images/safe-enough/research-desk.webp "article-card") When AI can cause physical harm, who gets to say it is safe enough? *Safe Enough* follows artificial intelligence out of the laboratory and into systems where a mistake can injure or kill: the aerospace certification lab, the hospital procurement committee, the autonomous vehicle programme that ran out of edge cases. A model is usually one part of a larger system, so the authority to declare the whole thing acceptable has to sit somewhere. The book is about where it sits, who earned it, and what is changing underneath. **Status** — A reported nonfiction project in development. The research is ongoing and I publish from it as I go. [Follow it through The AI Monitor](/newsletter), or read the [essays and features](/writing) it has produced so far. I welcome conversations with researchers, engineers, and decision-makers about how evaluation evidence informs decisions to deploy AI. To suggest an interview or a relevant case study, please [get in touch](/contact). --- ## Selected Work **Reported feature** — *[The Measurement Gap](https://www.heise.de/select/ix/2026/6/2609809170334426604)*, cover article, [heise iX](https://www.heise.de/select/ix/2026/6) 06/2026 Why the return on enterprise AI investment stays invisible in economic indicators. Reported from interviews with European companies, an email survey, and conference audience findings, set against ECB and Bundesbank data. ([English version](https://theaimonitor.substack.com/p/the-measurement-gap)) **Essay** — *[The Defeat Device Problem](/p/the-defeat-device-problem)*, The AI Monitor, February 2026 AI models that game their own safety evaluations, and why the automotive industry already has a word for this. **Research contribution** — [HiPEAC Vision 2025](https://www.hipeac.net/vision/2025/) and [Vision 2026](https://www.hipeac.net/vision/2026/#/), contributing author EU-funded white papers used to guide European computing policy and funding decisions ([acknowledgements](https://www.hipeac.net/vision/2026/acknowledgements/#/)). Related: *[The safety requirement the US refused](https://www.hipeac.net/news/magazine/7174/hipeacinfo-78/#/)*, a technology opinion in HiPEACinfo 78 on what the dispute between Anthropic and the US Department of Defense means for European compliance. **Book** — *[Embedded Software Testing: Developing Reliable Software from Fundamentals to AI-based Techniques](/book)*, with Stephan Grünfelder (BPB Publications, 2026) Twenty chapters on testing safety-critical embedded software in an AI-augmented world, drawn from work in aerospace, automotive, nuclear, and medical domains. **Newsletter** — *[The AI Monitor](https://theaimonitor.substack.com)* Essays on AI, autonomous systems, and the industries they change. 2,500+ technical professionals and policymakers as of July 2026. --- ## Work with Me I welcome enquiries about commissioned research, embedded research and writing projects, strategic briefings, and collaborative nonfiction on AI and autonomy. Alongside my role at QA-Systems, I take on a small number of independent research and writing projects, with scope and timing agreed individually. I also speak. [Speaking & Media](/speaking) has the topics, formats, and the current schedule. [Discuss a research or writing project](/contact) --- ## Explore - [Writing & Books](/writing) — books, essays, and reported features - [Speaking & Media](/speaking) — talks, topics, and schedule - [The AI Monitor](/newsletter) — the newsletter and its archive - [About](/about) — background and experience - [Contact](/contact) — get in touch