The scientist developed the basic framework for the new IEEE standard.
(Photo: David Payr)
Dusseldorf After six years of work, with 34 computer scientists from six continents, the time had finally come for AI researcher Sarah Spiekermann. On Tuesday, she presented the 83-page manual at the University of Vienna, which companies and software developers can use to build ethically justifiable technologies.
Together with the global engineering association IEEE, Spiekermann developed rules for software developers, companies and start-ups. With the help of this standard work, you should be able to develop digital technologies in such a way that they do not have any negative influences or effects on the people who use them.
The IEEE 7000, as the set of rules is called, is “the first globally applicable process standard”, says business IT specialist Spiekermann, who was in charge of creating the standard. “A key pillar is having control over data processing partners, especially AI partners,” she explains. “Another pillar is: to consider what consequences an AI has on people if they use it.”
In other words: like the warning notices on a pack of cigarettes, the consequences of digital products should also be clear. To date, there have been numerous cases in which technology such as artificial intelligence had undesirable side effects. The application bot from Amazon is known a few years ago, which hired women at a disadvantage because it was mainly fed with data about men.
Top jobs of the day
Find the best jobs now and
be notified by email.
Or stories from whistleblowers: Ex-Facebook employee Frances Haugen recently made public that Facebook had known for a long time through studies that its services could have psychological consequences for its users.
The EU is also working on finding a controlled way of dealing with artificial intelligence. “The Commission is currently reviewing the new IEEE standard,” confirms an EU Commission official. “We welcome the efforts of IEEE to integrate ethical considerations into the design of artificial intelligence.” The standard is in line with a proposed regulation on AI, which means creating specific rules for the design, development and use of AI applications it.
Human well-being should be in the foreground
Spiekermann, born in Düsseldorf, not only headed the working group, she had already developed the basic framework for the new IEEE standard. In English she calls it “Value-based Engineering”, that is, value-based construction technology, according to which IT development should primarily aim at human well-being.
The 48-year-old herself worked for technology companies, for example for Openwave Systems in the USA. She has already worked in various working groups on EU data protection regulations or for an OECD report on Big Data.
For around 14 years she researched the topic of “privacy vis-à-vis machines”, and since 2014 she has been researching ethics in IT system design. Since 2009 she has headed the Institute for Information Systems and Society at the Vienna University of Economics and Business. “At the moment it is still said that AI has to be privacy-friendly, transparent and reliable,” criticizes Spiekermann. “But these are requirements that you expect from a functioning system anyway, so that you as the buyer know what it is doing.”
She has now introduced into the IEEE 7000 standard what consequences the technology can have for society and for individuals. “For example, how technology can change a person’s character or virtues,” says Spiekermann.
The EU is also working on “digital principles”
Stefan Ullrich from the Ethics Section of the Society for Informatics has also already looked at the standard. “I can already imagine that development teams with this standard will be much more diverse and that the question of how we build software will also become more important.”
The EU is currently working on the so-called digital compass, which will implement the EU’s precise digital goals. One of them: By 2030, three out of four companies are to use cloud computing services, big data and artificial intelligence. For this purpose, “digital principles” should also be created, says a Commission official. “The preparatory work is still ongoing and the Commission will publish the draft declaration by the end of 2021.”
Regardless of whether Spiekermann’s set of rules will find its way into legislation, companies can purchase the standard from the engineering association. The business IT specialist also has other plans: “We are now starting to train a first generation of coaches and consultants who can carry the standard into the company.” In addition, says Spiekermann, the issue of licenses is already being considered.
More: Ethics in the digital age – why we need an organic seal for software