MIT Technology Review 4/23: How can we tame the new AI models?

Artificial intelligence went mainstream almost overnight with ChatGPT. Even if the initial hype will pass, the big language models like GPT-4 and the software world that is created around these generative AIs are so powerful that they will change our everyday lives. In the future, AI will simply do many tasks faster and better. This will have a major impact on research, education, marketing, journalism and much more.

Technology Review”,”Tierversuche”,”Wirtschaft”],”mpos”:[“understitial”,”top”],”themenhub”:”yes”}” type=”gpt” unit=”/6514/www.heise.de/tr/tr-inhalt” width=”300″/>

How intelligent the language models really are

Apart from the fact that language models still regularly hallucinate: do they have real intelligence? One could take the easy way out and say that of course they don’t understand what they write. But is it really that simple? Research teams use various tests to search for traces of intelligence.

How AI models are changing society

At the same time, the risk of abuse is great. The term “fake” takes on a whole new meaning when you look at the fake pictures of the pope in a white down jacket, for example, which can hardly be distinguished from reality. The author Eva Wolfangel looked into different areas of society and asked experts what effects large AI models will have on our working life, education and art.

Guidelines: How do I train an AI?

And that is just the beginning. Deep fakes, disinformation and attacks can cause confusion and manipulation with significantly less effort and a deceptively real presentation. In her text, author Katja Scherer asks how AI can be raised ethically in order to minimize such abuse – and provides answers.

Technology Review”,”Tierversuche”,”Wirtschaft”],”mpos”:[“3″],”themenhub”:”yes”}” type=”gpt” unit=”/6514/www.heise.de/tr/tr-inhalt” width=”300″/>

The fight for the AI ​​Act

And what does smart regulation look like? With the AI ​​Act, the EU is in the process of introducing a law that establishes comprehensive rules for AI – along their diverse applications and risk scenarios. A real fight has broken out between Big Tech lobbyists and the EU. Our author Bernd Müller gives an insight into the tug of war.

The genie is out of the bottle. We can’t catch him anymore. Showing him his limits and using his abilities to shape the future: that is the task now.

Other highlights of the issue

• Medicine: In the United States, the FDA no longer requires animal testing for drug testing. The end of the laboratory mouse?

Read MIT Technology Review here:

(lca)