What threatens if AI gets a body

Artificial intelligence is conquering our everyday lives in the form of software solutions. But digital assistants are just the beginning of the AI ​​revolution. In the next step, artificial intelligence will help us conquer our physical world and get a body. A commentary analysis.

What is Physical AI?

• The Gabler Business Lexicon defines Physical AI as “the connection of artificial intelligence with physical systems such as cameras, devices, systems, drones, robots and self-driving cars and buses. It enables these, to perceive complex processes in the physical worldanalyze, evaluate and execute.”
• So far, training new technologies such as humanoid robots or gripper arms in production has been very difficult and costly because there is little data and a lot of effort was required. Set at this point Multimodal Large Language Models (MLLMs).
• They train humanoid robots Billions of real data setswithout endangering the expensive technology. An illustrative example is Nvidia Cosmos. Physical AI teaches the autonomous car the dangers of everyday traffic through countless video recordings – without a real journey on the road, during which an accident could cause damage to the prototype worth millions.

Humanoid robots are still far too expensive

The Figure 03, Tesla’s Optimus and the robots from Boston Dynamics have one thing in common: They sometimes deliver impressive demonstrations, but are Too expensive for use in your own four walls. A good example of this is the care robot Moya, which is said to cost just under 180,000 US dollars.

Physical AI could become a game changer for humanoid robots. Through the uncomplicated training and the Humanization of machines By dealing with physical laws and unpredictable human environments, our robot assistants are getting better and better. This in turn enables scaling in the industry. There the robots can in one learn in a protected environment and get even better.

As soon as humanoid robots work in factories, robot producers like Elon Musk’s Tesla can Manufacturing and development costs reduce significantly. This in turn makes it possible to move into our private lives in the future.

Voices

• Markus Hacker, Director Enterprise DACH at Nvidiaoutlines the further development of robotics in a guest article: “Robots will become intelligent units that not only perceive, reason and plan, but can also act independently and learn from their experiences. Their areas of application will extend from operating rooms and data centers to factories, traffic control systems and entire cities. Static, manually controlled systems are increasingly giving way to interactive, intelligent solutions.”
• Physical AI not only has an impact on our gross domestic product, but also on our world of work. Point to that Theresa Wolf and Andrea Willige from the World Economic Forum in a paper: “Machine operators will become robot technicians, logistics teams will coordinate mobile robots, maintenance teams will shift their focus to predictive maintenance, and manufacturing engineers will focus on training and optimizing AI and robotics systems.”
• The professor contradicts any possible doomsday scenarios for humanity Wolfgang Wahlster, scientific director at the German Research Center for Artificial Intelligencein an interview: “Machine learning is fundamentally different from human learning. AI is by no means about perfectly replicating human behavior with all its strengths but also its many weaknesses, but rather we want to assist people with AI where this makes sense. When it comes to learning regularities in very large amounts of data, machine learning methods are already superior to humans today.”

Physical AI: castle in the air or everyday helper?

Physical AI is comparable to the Metaverse at this point. A technology that offers great potential, but is ahead of its time and therefore partly on Castles in the air in fantasy worlds builds. Realistic estimates suggest that the first humanoid robots will appear in factories towards the end of the 2020s.

They could help in our household from the mid-2030s. So despite Physical AI’s steep learning curve, we’re talking about a period of at least a decade – a period that can hardly be reliably predicted.

What is certain, however, is that the success of Physical AI could help to relieve physical strain on people. When we do less manual work, we can focus on more important tasks.

This includes, for example, caring for relatives or spending time with the family. Perhaps technological progress will actually provide economic relief for humanity and a social and interpersonal revolution.

Also interesting: