THE NEWS

🤖The EU agrees on AI rules

The EU is setting up its first-ever rules for AI. The goal? Make sure AI is fair, safe, and eco-friendly, without stopping all the AI innovation. It's like they're trying to domesticate a T-Rex – ambitious, a bit scary, but definitely necessary in a world where T-Rexes are popping up everywhere.

Here’s the Lowdown on the New Rules:

No to Dangerous AI: First off, the EU is saying a big ‘no’ to AI that can make us do things against our will. Imagine a toy that could convince kids to do dangerous stuff – that’s a no-go. And remote-controlled systems with real-time facial recognition, like drones that can recognize your face and follow you? Also not allowed.

But, there’s a catch – the military and police can still use the latter. It’s a shame that the EU watered down their originally proposed unconditional ban on facial recognition for law enforcement purposes.

Yes, But Be Careful: For AI that’s kind of risky, like self-driving cars or medical tech, the EU will require a review before they are approved. This also includes border surveillance, migration control, police work as well as AI for HR and law purposes.

Keeping an Eye on Chatbots: The EU deemed generative AI not too risky, but wants the companies behind these AI systems like chatbots (i.e ChatGPT) and image generators, to come clean about how they work.

They must also tell the EU how the AI was trained and which copyright-protected data was used. This will have huge implications, since these AI companies have been obviously training on copyrighted material but kept quiet about it, and haven’t compensated any of the artists or IP owners for their work.

The EU also wants all content generated with these AI systems to be labeled.

Just a Heads-Up for the Simple Stuff: Interestingly, the EU is not sweating the deepfake videos or AI customer service. Companies just need to tell us we’re chatting with a bot, not a real person.

We’ll have to wait longer for these regulations to kick in than GTA VI, though. All these new rules should get the green light in 2024, and then countries in the EU have two more years to get everything in place.

That’s just crazy. By the time these rules are implemented, they might already be out of date. AI moves at a break-neck speed and the world will be a different place in two and a half years.

The European Union is really stepping into new territory here. They’re trying to make sure that as AI grows and gets more into our lives, it stays on the good side, respects our rights, and doesn't go rogue. It's a big deal and could set the trend for how the rest of the world handles AI. It pushes these AI companies in the right direction on many fronts, but implementing seems slow and I’m not sure it won’t stifle European innovation.

ONE MORE THING

Real-time AI conversations are here

I know you guys like clicking on links like cats like water, but I hope you watch the video in this tweet, because the speed of response of this AI is insane. AI call agents are here.

⌚ If you have one more minute

• 🇫🇷French AI start-up Mistral secures €2bn valuation

• 🤖Mistral released Mixtral 8x7B that outperforms Llama 2 70B on most benchmarks with 6x faster inference, works in 5 languages.

• 🤖 Most Businesses Have Yet To Harness AI

• 🎥 MotionCtrl: A new video generation tool with flexible motion controller

• 💭 Andrje Karpathy on the ‘hallucination problem’ of LLMs

AI Art of the Day 🎨

Now, this is what I call the real Miss Universe competition. This hottie is from Saturn. Check out all the contenders here. Made with Midjourney by u/Hysteria54.

🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄

That's all for today, folks!

If you enjoyed this, please consider sharing this hand-crafted newsletter with a friend.