How tech firms are trying to make ethical AI: a few are formalizing ethics processes, Facebook has an automatic adviser to spot potential machine learning bias
Tom Simonite / Wired : Tweets: @wired and @jtasioulas Tweets: @wired : New research institutes, industry groups, and philanthropic programs have sprung up to help tech companies think through the ethical challenges raised by AI http://www.wired.com/... John Tasioulas / @jtasioulas : “There's a lot of high-falutin talk but everything I've seen so far is naive in execution” - @WendellWallach on corporate AI ethics boards http://www.wired.com/...
Context & Ripple Effects
This 2018 piece is an early snapshot of what became the industry's standard playbook: tech firms formalizing ethics processes, with Facebook going furthest by embedding an automatic adviser in its machine-learning pipeline to flag potential bias rather than relying on a human review board alone. At the time, the infrastructure was new — research institutes, industry groups, and philanthropic programs were just forming to help companies think through AI's ethical challenges.
The subsequent coverage shows how contested that playbook became. Critics charged that ethics boards and principles functioned as a strategy for avoiding government regulation, while commentators argued boards can raise awareness but cannot implement good outcomes. By 2020 the pattern had matured enough that Google planned to sell AI ethics advice — bias-spotting and project guidance — as a commercial service.
First-order effects
- Facebook shifts bias detection from periodic human review to continuous automated checking inside its ML systems, making ethics a pipeline feature rather than a committee function.
- Firms adopting formalized ethics processes gain a public-facing answer to AI criticism, though Wendell Wallach's assessment in the coverage — high-falutin talk, naive execution — frames the credibility risk they carry.
Second-order effects
- Google's move to package ethics advice as a paid service turns what Facebook built internally into a product category, with large platforms positioned to sell governance tooling to smaller firms.
- Smaller players like Clarifai push back on the model entirely — its CEO argued internal ethics officers do not suit small companies — creating a market split between firms that institutionalize oversight and those that skip it.
Third-order effects
- If self-governance stays thin, oversight defaults to whoever is adjacent to the work — the coverage notes researchers worrying that ethical review falls to peer reviewers, a gap other technical fields solved with dedicated institutions.
- The pattern points toward either genuine operational governance embedded in ML tooling or externally imposed rules; the corpus shows both forces growing, with critics explicitly framing corporate ethics structures as an alternative to regulation.
The trend: AI ethics is migrating from advisory boards toward embedded tooling and commercialized services, while the question of whether industry self-governance can substitute for regulation remains unresolved.