✅ ChatGPT can now browse the web to answer your questions with the arrival of third-party plugins (see more)

🎈 This week in the AI world:

• Google launches Bard.

• Adobe announces its Firefly image creator.

• Microsoft unveils Bing Image Creator.

• NVIDIA launches cloud tools for generative AI.

• Opera launches its LLM-based browser.

• GitHub unveils CopilotX.

• Canva releases a big update full of AI features.

• Unity announces an AI feature to generate 3D scenes from text.

• RunwayML announces Gen-2 (Text-to-video)

✅ Now on Chinese television, instead of a real reporter, you can see an AI bot

According to representatives of the TV company, this was the best measure to improve the quality of broadcasts and increase production efficiency.

At this rate, half of the people on the planet will be out of work

✅ BREAKING: A petition is circulating to PAUSE all major AI developments.

e.g. No more ChatGPT upgrades & many others.

Signed by Elon Musk, Steve Wozniak, Stability AI CEO & 1000s of other tech leaders.

They are all very concerned about the capabilities of GPT-4: in their opinion, if an even more capable AI emerges, it could pose a real threat to humanity. Within these six months, humanity should be able to adapt and adjust to the changing influence of neural networks.

🎈 OpenAI is already preparing GPT-5 according to reports, it is expected to be released by December of this year.

In addition, GPT-4.5, an intermediate version between generations, is set to be introduced in September-October.

GPT-5 is expected to be trained by the end of the year and promises a revolution in the field of AI. It is even being compared to the launch of a new internet (as reported by OpenAI developer Siqi Chen)

🔴 The AI control problem explained

The idea of the AI control problem is that AI will eventually become better at making decisions than humans. According to this theory, if we don't put the right preventive measures in place, we won't have the opportunity to correct things later, thus letting the AI take control.

However, AI and machine learning systems have limitations, being bound by the laws of physics, computational complexity and the processing power of the devices that support them. Despite this, it can be assumed that these limits are far beyond human capabilities. Thus, superintelligent AI systems can pose a major threat if they are not properly designed with safety features to control any potentially dangerous behavior.