Amazon founder Jeff Bezos is the latest tech billionaire to come out of semi-retirement and return to the frontlines of the artificial intelligence (AI) boom that has fueled a new generation of startups.Bezos, who is among the world’s richest people, will be leading an AI startup called Project Prometheus as co-CEO, along with former Google researcher Vik Bajaj, according to a report by The New York Times. Project Prometheus has already raised $6.2 billion in funding, partly from Bezos, making it one of the most well-financed early-stage startups in the world.The AI startup will reportedly focus on applying AI to physical tasks, including robotics, drug design, and scientific discovery.While Bezos has continued to focus on his private spaceflight company and SpaceX rival, Blue Origin, this is the first time he has taken on a formal operational role in a company since stepping down as CEO of Amazon in 2021. His latest venture comes amid an increasingly crowded AI market comprising hyperscalers such as Google, Meta, and Microsoft, pioneering companies like OpenAI and Anthropic, and other upstarts that are carving out niches to gain a competitive edge such as Thinking Machines, Cohere, and Scale AI, among others.What is Project Prometheus? Who is part of it?Named after a popular figure in Greek mythology who stole fire from the gods and gave it humankind, Project Prometheus reportedly intends to build AI systems that are capable of accelerating scientific discoveries in fields ranging from engineering and manufacturing to computers, aerospace, and automobiles.The AI startup has already assembled a 100-member team comprising researchers that have reportedly been poached from leading AI research labs such as OpenAI, Google DeepMind, and Meta.However, Project Prometheus remains largely under wraps with no formal announcement yet and key details such as its base of operations still unclear.Story continues below this adThe other person heading up the AI startup besides Bezos is Vik Bajaj. In the past, Bajaj has worked closely with Google co-founder Sergey Brin on projects under Google X, dubbed as ‘The Moonshot Factory’ since it focused on building ambitious products such as Wing, a drone delivery service. Waymo, the self-driving car company, is also a result of this research effort.What sets Project Prometheus apart? Who are its competitors?Project Prometheus is reportedly among the fresh crop of AI startups and research labs that are looking beyond large language models (LLMs), the technology powering AI chatbots like ChatGPT. Instead, these startups are focusing on building ‘world models’ or AI systems capable of learning from the physical world.Periodic Labs is one such startup that has been founded by several prominent researchers who left Meta, OpenAI, and Google DeepMind to develop technology that can accelerate new discoveries in areas like physics and chemistry. It plans to build its own lab in California, United States, where robots will run scientific experiments on an enormous scale.Another startup working in this domain is Physical Intelligence. Its investors include Jeff Bezos, Bond, Khosla Ventures, Lux Capital, OpenAI, Redpoint Ventures, Sequoia Capital, and Thrive Capital, as per the company’s official website.Story continues below this adThese companies resonate with the view of Yann LeCun, Meta’s chief AI scientist and a professor at New York University, who has repeatedly said that LLMs are a “dead-end” and that they will never be able to achieve human-level reasoning. The AI/ML pioneer is reportedly planning to leave Meta and set up his own venture that will focus on world models.“If I tell you ‘imagine a cube floating in the air in front of you. Okay now rotate this cube by 90 degrees around a vertical axis. What does it look like?’ It’s very easy for you to kind of have this mental model of a cube rotating,” LeCun has been quoted as saying by Gizmodo.He has said that while LLMs are trained on massive amounts of data scraped from the internet and other sources, they have not been derived from processing sensory data about the physical world. As a result, LLMs can generate coherent text but lack true understanding, planning abilities, and meaningful interaction with the physical world, making their intelligence far below even that of a housecat, according to LeCun.