MIT’s Kenneth Oye urges focus on AI data governance over sci‑fi AGI fears

Kenneth Oye’s interview is a useful counterweight to the AGI‑doom vs. AGI‑utopia framing that dominates much of today’s discourse. From his vantage point at MIT’s Program on Emerging Technologies, the near‑term dangers are less about a single runaway intelligence and more about how AI is woven into defense, credit scoring, sentencing, hiring and admissions decisions—domains where biased data and opaque models can quietly entrench or amplify injustice. ([elpais.com](https://elpais.com/proyecto-tendencias/2026-06-29/kenneth-oye-profesor-de-mit-no-me-preocupa-tanto-eso-de-la-inteligencia-general-que-va-a-dominar-el-mundo.html))

For the race to AGI, this perspective matters because it shifts the center of gravity from speculative end states to concrete governance of data and power. Oye points out that privately controlled datasets are becoming the real choke point: whoever owns and curates these corpora wields disproportionate influence over how models see the world. That’s as true for future AGI‑class systems as it is for today’s credit‑risk models. ([elpais.com](https://elpais.com/proyecto-tendencias/2026-06-29/kenneth-oye-profesor-de-mit-no-me-preocupa-tanto-eso-de-la-inteligencia-general-que-va-a-dominar-el-mundo.html)) In practice, his argument suggests we need regulatory and institutional innovation around data trusts, auditing, and access regimes at least as urgently as we need new techniques in alignment or interpretability. If we can’t get data governance right at GPT‑4–level capability, we’re unlikely to manage it gracefully as systems grow more general.