What has changed in ten years

The IEEE Standards Association is marking the tenth anniversary of the first Ethically Aligned Design (EAD) document, the report published under the Global Initiative for Ethical Considerations in Autonomous and Intelligent Systems, launched in 2016. According to IEEE Spectrum's account, the stated goal was to shift ethical attention from the stage after autonomous and intelligent systems are released to the stage of their design. The report identified eight foundational principles, including human rights, people's well-being and transparency.

The source reports that more than one hundred AI ethics experts contributed to the first version and that the second version arrived in 2017. John C. Havens, founding executive director of the initiative and now global director of IEEE Planet Positive 2030, says that both versions gathered hundreds of pages of public comments and that by 2019 contributors exceeded 400. He is also credited with the statement, reported by the newspaper, that EAD was the first document of its kind because it was not limited to a list of principles and came from an organization with IEEE's credibility.

From principles to standards and certifications

The most significant shift concerns translating principles into operational tools. Again according to the source, the IEEE 7000 standards series grew out of the EAD framework: it currently comprises 29 projects, 11 of which are published, with others in the process of being added. The topics covered range from the process model for ethical system design (IEEE 7000-2021) to transparency of AI systems (IEEE 7001-2021), from data privacy (IEEE 7002-2022) to considerations of algorithmic bias (IEEE 7003-2024), through to impact on people's well-being (IEEE 7010-2020), machine-readable privacy requirements (IEEE 7012-2025) and systems that simulate human empathy (IEEE 7014-2024).

On the verification front, IEEE SA offers AI ethics programs including CertifAIEd, with two certifications: one for people and one for products. The first trains individuals capable of evaluating autonomous and intelligent systems against IEEE ethical methodology and criteria; the second assesses whether an organization's tool or system meets the criteria of IEEE ontological specifications, with an assessor who examines the product and a conformity assessment program that can issue the mark. The Responsible AI for Children program has also been introduced, certifying children's products independently tested and verified against design and ethics requirements.

The network and what remains to be seen

The collaborative dimension has grown: the AI Systems for Global Flourishing Collaborative Industry Collections group, the source reports, has more than 200 members and is working on a systemic framework to translate the concept of flourishing into methods, metrics and tools applicable throughout the systems lifecycle. Havens explicitly links this work to the well-being chapter of EAD and to IEEE 7010, arguing that if human well-being and planetary sustainability are not addressed at the design stage, they remain peripheral concerns.

For the tenth anniversary, IEEE SA launched a website, a blog and a five-episode podcast series, launched on October 21 with releases every two weeks, devoted to the EAD vision, the choice of principles and their translation into the 7000 standards.

The material, however, is a celebratory account produced by the organization itself: it contains no independent assessments of the standards' effectiveness or data on companies' actual adoption of the certifications. Havens's statements about EAD's pioneering nature are attributed and unverified; it therefore remains to be established, using third-party sources, how much the framework has affected development practices and not merely documentary output.

Source and attribution: this original summary was prepared by AICronaca based on the report by IEEE Spectrum — AI ↗The statements are attributed to the source; any information limitations: The text is based on an IEEE Spectrum article that reflects IEEE SA's perspective; no external assessments, certification adoption data or independent verifications of the standards' impact are present. Havens's claims are reported as statements..