If you dream of a day, artificial intelligence reaches a person’s abilities Airbus CSO Paul Eremenko says he always did it in the context of the construction of machines in the real world. “I want AI warden who can create star ships and dysana,” he said Wealth-And the reflection-hypothetical sci-fi mega structure that would use the energy from the star.
While his dream is still far, Eremenko puts the base. He united his strength with the former Google Deep mass Researcher Alex Gordic and Adam Nagel, an engineering leader, previously at the Acubed Innovation Center, the Innovative Center for Airbus. Together, they founded the P-1 AI, which today appeared with Stealth with $ 23 million, led by radical enterprises. Other investors include Global Global, schematic enterprises and lerer hippos, as well as well -known angels such as Google Deepmind Chief Scientist Jeff Ding and Vice Prime Minister, Peter Valder.
P-1, named after Adolescence age p-1The science fantastic novel of 1977, Thomas Joseph Ryan about the Sentient AI, develops an engineering assistant under the name Archi. Similar to other Assistant II, such as AI-Coding Devin from Cognition AI, the idea is to embed the Archi as a junior member of each engineering team-for repeating, but time that causes time such as interpretation of requirements, creating early design concepts and conducting rules. This is an early step to a much more ambitious vision: the use of AI for the end to develop the complex machines of the future.
Eremenko said he was surprised that no one was already working on this purpose, but he quickly found out why. Just like in self -governing cars and robots, to train AI to build machines requires a large amount of learning data. The key, he explained, is the modeling of realistic engineering systems, creating virtual models of real components such as engines, pipes and shaft. Then these physics simulations are combined in different configurations to create data that are then used to train AI models that help automate engineering design.
According to Gordic, it looks like Google Deepmind used games to help train Alphago, AI, which defeated Human Champions in Go, a gloriously sophisticated board game. “Alpha was prepared initially by imitation data from real players,” he said Wealth. Now it will train and refine large language models (LLMS) and other AI systems to understand and change sophisticated engineering projects in rich physical systems such as cooling data centers or VVC systems.
In order to go beyond the “glorified auto -filing” capabilities such as Chatgpt, he explained, models must be useful for engineering tasks. Therefore, AI must actually understand commands and follow the instructions. The powerful combination of AI models, which are trained on synthetic data built on physics modeling, and can then understand and act on this data, makes truly automated engineering aid by reality. “We train Archi on synthetic data to force him to the college engineering level,” Eremenko continued. But after the Archi dislocation, you can learn from human reviews and real data from companies using AI.
P-1 investors, Eremenko said, interested in more substantiated short-term startup plans, but they are especially excited about the future. “Many of us in engineering and AI World we grew up on science fiction, and sci-fantastic promised us a super intelligence that will build star ships,” he explained.
Big executives like Autodesk. Siemens and IBM Working on the use of AI for engineering, but they do not create a new class of Generalist Engineering AI assistants, and they do not go after the same grand vision of machines built by AI.
However, Eremenko and Gordic insist that their very realistic and purposeful path, and this is not just a research project with an uncertain term. “We’re not going to be a 10-year moonshine,” Eremenko said. “This is a very pragmatic deployment and the way to the market.”
Originally this story was presented on Fortune.com
https://fortune.com/img-assets/wp-content/uploads/2025/04/P-1-Team_Triptych_Color-1.jpg?resize=1200,600
2025-04-28 13:00:00
Sharon Goldman
