AI and Democracy: Vielleicht nicht wahr ist auch schon schlimm

Read the KI newsletter “Natürlich intelligent” from September 5, 2024. In the Newsletter you can read Donnerstag per Mail zu erhalten, report Sie sich here an.

Is that so, was it really true? Or was it a mistake?

This Frage is not new. Imagine it is photography. This is Photoshop gibt. If you have a smart intelligence in professional photo industry, you can read more.

In the verbs you see in German: Die Sorge, dass durch KI de Grenzen zwischen Realität und Fälschung zunehmend verschwimmen, ist durchhaus berechtigt. Deepfakes, generated by the AI ​​of the Freiburger Start-ups Black Forest Labs, whistled on the Internet in August: Joe Biden in Windeln, Elon Musk in a Nazi uniform, a swaggering Kamala Harris.

Then Google launched the Pixel 9, a smartphone with AI functionality, which with just a few clicks can display grim images: a few white people have a legend, a huge explosion of explosive animals.

Beeindruckend daran: Das Light und de Größenverhältnisse auf de KI-generierten Bildern stimmen meist, sid gestochen scharf. When a person selects a finger from the hand. Fake-Bilder von der Realität zu unterscheiden, konnte in Zukunft also become ever more shy.

Experts warn that this will happen in the long run, but generative AI will become easier, mass disinformation will be set up. There are people who look after their interests and the politics of private individuals who make discrediting possible.

US presidential candidate Donald Trump has sketched the first picture, the vermeintliche images that singer Taylor Swift and his fans have received. It was an undisputed matter that a democracy had done, when the political enterprise itself took on its political activities to let the images appear.

But the whole thing is the Sache nicht.

There are only a few persons, groups or states, with KI Wahlkämpfe zu beeinfussen. It is an example of Indonesia being Slovak in the US. But study the results here: Bisher shinen die Bemühungen nicht sonderlich erolgreich zu sein.

As one of the more than a hundred national Wahlen in diesem and vergangen Jahr stellten Forscher des Alan-Turing-Instituts aus Großbritannien fest: Bei nur 19 von ihnen fand a Beeinflussung durch KI statt. Moreover, the Forscher has put an end to the disappearance of the tatsächlichen Wahlergebnissen and the erwarteten Abschneiden aufgrund von Umfragedaten gegeben. The information comes from the Schluss: “The evaluation of the knowledge within the Wahlergebnisse since it is limited.”

Was it a passion, if everything was going so well, was it on the internet that now a theoretical problem was possible?

Derzeitige Fortschritt in der KI kan een Wendepunkt darstelling, such a basic legend changes, was a wem vertrauen.

In a world, in everything possible, you can get an image, a video, every voice – in that world it is a bit, the reality of life. All information, the AI ​​recording in the display of your message, the display of real images, videos and audio data is not authentic. Everyone can say: “That’s what an AI made, that’s fake” – and that makes the responsibility easier to see. Scientists call these phenomena the “Liar’s Dividend” (in German something like: “The profit of the liar”). More normal deepfakes and other AI photos were discovered through a sale of information and forgeries. Everything is authentic.

There is no question of a pure Theory, but of a Beispiel of Mitte August. At a Wahlkampfveranstaltung at the airport in Detroit a beautiful jubilant person of the American presidential candidate Kamala Harris. Take pictures and pictures of people during the Blickwinkeln-fest.

When Donald Trump came to power, he had only just started using a photo of a photo of Harris’s airplane that he saw when he proposed AI. “Da war niemand beim Flugzeug”, Trump wrote on the Platform Truth Social. Harris generated the image AI.

People and states with bad intentions are used for their purposes. It is difficult to imagine that you can use technology for democracy and democracy.

If you want to spread the technology, this is hard. The blind fear of AI can be caused. If a democracy is a fact and a common reality, and all glauben, the years of frost are over.

Links to more information

• In welchem ​​​​Ausmaß could KI beeinfussen those US-Wahlen? Forscher for the US-American Brookings Institution das Für en Wider: Misunderstood Mechanics: How AI, TikTok, and the Liar’s Dividend Could Sway the 2024 Election
• What wrong answers from chatbots on Wahlen and our warnings are possible, but think about it Atlantic Ocean-Author Matteo Wong to: Chatbots Are Poised to Twist Reality (The Atlantic)
• Generative AI is not in Photoshop – This is an argument made by journalist Jess Weatherbed: Hi, you’re here because you said AI image editing is like Photoshop (The Verge)

Thinking about AI

• Columnist Kevin Roose has looked at the mind-boggling chatbots of his life – and said of a new industry: How Do You Change a Chatbot’s Mind? (The New York Times)
• AI can perform the internet listings. There are companies active with OpenAI, Perplexity and Google itself. That put Medienhäuser in a dilemma: Google’s AI search engine gives sites a desperate choice: share data or perish (Bloomberg)
• Whoever encounters large language models in their answers is not clear in English. This is the Fortschritte: Researchers discover how large language models work (The Economist)

Ruminating with AI

• At Fal.ai, the AI ​​image generator Flux from Black Forest Labs can be tested, as the best in the world: https://fal.ai/models/fal-ai/flux-pro
• Conversation with Google – go to Gemini Live on an Android device and Google’s AI subscription: https://support.google.com/gemini/answer/15274899?hl=de
• Now that we have found OpenAI’s new AI machine, Google’s data can be used. Die Washington Post with my testimonials and first impressions: Few Have Tried OpenAI’s Google Killer. Here’s What They Think. (The Washington Post)