döbeli honegger prognostiziert, infolge der neuen tools werde die informationsflut ansteigen. auch die flut an fakenews dürfte zunehmen.

Döbeli Honegger sagt, es werde wichtiger denn je, Schülerinnen und Schülern Medienkompetenz beizubringen. ChatGPT und alle anderen Text- Roboter, die jetzt in der Pipeline sind, werden zu einer Flut an geschriebenen Informationen führen. Fake News werden immer seriöser daherkommen. In der Schule müsse man lernen, wie ein KI-Text als solcher zu erkennen sei. Das müsse «Teil der Allgemeinbildung» werden, sagt der Dozent.

Indeed, new state-of-the-art artificial intelligence (AI) technologies can now generate text that mimics the style and substance of real news stories while overcoming the bandwidth limits that face human-generated text. If these models enable malicious actors to generate and publish crediblesounding news stories at scale, then the prospect for misinformation, defined as “false or misleading information,” 3 is high: the volume of inauthentic media could balloon, and the ease of text synthesis might further enable the coordinated hyper-targeting of articles to individual groups.

We found that people can be manipulated by AI-generated text such that they cannot discern real from synthetic content. The narrower consequence of a manipulable public is that misinformation campaigns have a ripe target. Malicious actors can easily produce AI-generated content and generate confusion about the truth, undermining trust in democratic institutions such as the media. More generally, however, the ease of manipulation suggests avenues for misinformation not in service of political persuasion but instead in sowing confusion and distrust. Following Kant’s belief in “a common sense as the necessary condition of the universal communicability of our knowledge," the erosion of common reference points has the potential to undermine the basis of coherent public policy.

AI’s ability to generate deep fake videos and new stories doesn’t frighten meeither. It just puts us back where we were when we used to get our newsfrom printed text. Anyone can put anything into the same font, print it out,and pass it around. The veracity of a news story has nothing to do with itsproduction value and everything to do with the integrity of the institutionproducing it. That’s why news organizations are supposed to work within asystem of “journalistic standards and practices.” They teach this stuff injournalism school, and places like BBC (and sometimes CNN) employ it. Yourwillingness to consider the ideas in this piece is dependent on your trust inme, a human being with a reputation at stake and, hopefully, a track recordof at least attempting to write truthfully.

A third category of risk involves bad actors taking advantage of the ability of large LMs to produce large quantities of seemingly coherent texts on specific topics on demand in cases where those deploying the LM have no investment in the truth of the generated text. These include prosaic cases, such as services set up to ‘automatically’ write term papers or interact on social media,23 as well as use cases connected to promoting extremism. For example, McGuffie and Newhouse [80] show how GPT-3 could be used to generate text in the persona of a conspiracy theorist, which in turn could be used to populate extremist recruitment message boards. This would give such groups a cheap way to boost recruitment by making human targets feel like they were among many like-minded people.