Greens: AI for climate and diversity, but not for killer robots


The parliamentary group of the Greens wants to show with its own strategy for artificial intelligence (AI) how this technology “can be used for the good of people and the environment”. In order to cope with the crises of the 21st century, modern means have to be used, they explain. AI should be used “for climate and health”. At the same time, it is important to prevent abuse through surveillance and disinformation: “It’s not about terminator, but about more efficient wind power and better cancer detection”.

AI is changing the life of society as part of the digital revolution “as fundamentally as the first agriculture in the Stone Age and industrialization did 200 years ago,” writes the opposition faction in its am Decision published on Friday. It is therefore important to establish clear framework conditions for the application of the technology. In the case of the federal government’s AI strategy, claims and reality are far apart.

“We align AI with sustainability and promote their commitment to climate protection,” the Greens pledge. Machine learning can accelerate the expansion of renewable energies and reduce costs by helping to select cheaper and better locations for wind and solar systems. These could be controlled and utilized more efficiently using short-term forecasts. In particular, AI can predict supply and demand in the electricity market very well. This makes it easier to reduce network charges and transaction costs.

The parliamentary group also wants to campaign “for targeted funding and standards for energy-efficient AI models and climate-neutral data centers”. Climate-damaging rebound effects would have to be taken into account from the start.

According to the Greens, AI is able to “make mobility safer, more environmentally and climate-friendly, more efficient and more convenient” and to build and control holistic transport systems. In the organization of traffic, however, the needs of autonomous cars should not be prioritized over those of bicycle and pedestrian traffic. Self-driving cars would have to “adapt to the mobility requirements of the environmental association”. It goes on to say: “For the Vision Zero of zero traffic fatalities, we want to use AI sensibly in the traffic turnaround.”

In the healthcare sector, MPs are planning to use AI to improve care and ensure that patient sovereignty remains intact. You advocate a decentralized research data infrastructure in contrast to the central one promoted by Federal Health Minister Jens Spahn (CDU). Test subjects and patients should be able to release information voluntarily and independently for specific or global research purposes and revoke it at any time. Such “standardized data” would have to be “pseudonymized and, if possible, anonymized and processed”.

“We are promoting the international outlawing of deadly autonomous weapons,” is a promise. Killer robots, “which are not subject to any control or control by humans when selecting and fighting targets,” represent “an unpredictable threat” and violate international law. On the other hand, there must be “common rules, international agreements, codes of conduct and stronger parliamentary control”. Germany should plead for a global export break here.

The Greens also want to draw “clear boundaries” for automated surveillance in public spaces. The red line is here with biometric systems for automated face recognition, for example. These represent a threat not only for human rights defenders and media workers in authoritarian states. In this country too, there are “significant risks for the protection of fundamental rights”. That is why the technology should be subject to a moratorium.

In general, algorithms should be created on the basis of democratic values ​​such as non-discrimination, fairness, transparency and accountability as well as “while respecting the right to self-determination”. Human control must have priority.

In order to prevent systematic discrimination as early as possible during the development of AI, the parliamentary group advocates incorporating diversity directly into technology using “Diversity by Design”. You have to rely on different disciplines, mixed teams and data sets.

The Greens are also pushing for binding guidelines for the use of AI in social networks and by the public sector. Automatic filters are particularly on platforms that create media publicity, “not in accordance with the fundamental rights protected freedom of information and freedom of expression”. The comprehensive tracking of users by platform providers for advertising purposes (microtargeting) should be replaced by “more socially acceptable systems” such as context-based advertising.

AI made in Europe really has to become a brand, the group demands a European-oriented research landscape. This is the only way for the EU to be “economically sovereign” in this area of ​​technology. From the many initiatives for a European Open Science Cloud and Gaia-X, for example, concrete and better data availability for research based on the so-called fair criteria (findable, accessible, interoperable, reusable) must result in a timely manner. A high level of IT security must be guaranteed right from the start. Networked data rooms and open standards are needed to give medium-sized companies access to large amounts of data.


To home page