After a good two years of work, the Enquete Commission on Artificial Intelligence (KI) of the Bundestag handed its final report to Parliament President Wolfgang Schäuble (CDU) on Wednesday. The 19 MPs and 19 external experts have placed the work under the model of a “human-centered AI” and are thus following in the footsteps of the Federal government, which with its AI strategy put the common good in the foreground.

Shaping the change in AI values ​​democratically The roughly 800 Pages run through a “yes, but” attitude towards the agreed next level of “digitization driven by technological progress”. The change in values ​​driven by AI is “not bad per se”, but it must be designed democratically, they say. AI applications should “primarily focus on the well-being and dignity of people and bring benefits to society”. This is the only way to exploit the positive potential of the key technology and to justify and reinforce the user trust required for its economic success.

The committee advocates AI made in Germany with “European characteristics” stands out from the state directional Chinese and market-driven US approach. However, the majority of those involved did not want to change the path taken by the EU Commission and the Data Ethics Commission of dividing AI systems into risk classes and possibly prohibiting particularly dangerous variants such as automated face recognition.

2017 the SPD had called for an algorithm TÜV so that no one would be discriminated against or harmed by software-based decisions. There is still “sympathy” for such regulatory requirements in the social democracy, explained the chairman of the AI ​​commission, Daniela Kolbe. But there was “no broad consensus” for this.

Prohibition of autonomous weapons “We will not come up with any DIN regulations”, um To contain KI, stated the SPD MP. But she thinks it would be good if companies listed possible risks to their system on a digital DIN A4 note. The majority agreed that AI solutions used by the state should also be checked for risks. The outlawing of autonomous, lethal weapon systems, which has already been called for in an interim report, is part of the final advice. recently failed to find a majority in the Bundestag. Nevertheless, SPD chairman Rene Röspel emphasized that the representatives expected the executive to “adopt at least parts of the recommendations” and, for example, incorporate them into the next versions of their AI strategy. It is also clear to the Social Democrats that new auditors are necessary. Just setting general standards and otherwise hoping for self-regulation of the economy is not enough.

“We looked at many practical examples and found: We do not need an algorithm TÜV and no new supervisory authorities”, said the chairwoman of the CDU / CSU parliamentary group, Ronja Kemmer. Politicians could “build on existing mechanisms and principles, such as the approval of AI systems or the protection of possible risks”. Refusing to use AI as a technology is “not a serious option”.

Does AI endanger jobs? The area was also controversial Work where the ambivalence of technology was clearly evident. AI opens up opportunities for employees, but also triggers fears in them, the responsible project group noted. Dangerous, physically difficult and recurring work could be reduced and AI systems could play a supporting role in solving complex tasks. On the other hand, there is “a fine line between supporting human activities and ways of restricting decision-making autonomy”. The consequences could be “work intensification”, more rigid control of performance and “devaluation of human experience-based knowledge”.

A key question for many is “whether one’s own employment relationship is endangered by the use of AI systems “, is d