Eric Zelikman, a prominent AI researcher who left xAI in September, is raising $1 billion for his new startup, Humans&, at a $4 billion valuation, according to sources familiar with the deal. The funding round is still underway and could change, but it underscores the growing trend of massive early-stage investments in AI ventures led by top-tier researchers. Zelikman, a Ph.D. candidate in computer science at Stanford, gained recognition as the lead author of a groundbreaking paper last year that explored how language models can be trained to "think before they speak"—a step toward more deliberate and context-aware AI responses. Prior to joining xAI in 2024, he worked as a machine learning intern at Microsoft and held a deep learning engineering role at Lazard. On the No Priors Podcast, Zelikman shared his vision for Humans&, a company focused on building AI models with emotional intelligence and deeper human understanding. He argued that current language models are too rigid and detached, treating each conversation turn as isolated rather than part of a continuous, meaningful interaction. “The models don’t understand the long-term implications of what they say,” he said. “They don’t really care about the person on the other end.” Zelikman believes the field has been misaligned, with too much focus on raw performance and not enough on collaboration, empathy, and long-term user well-being. “There’s so much talent out there, and I’ve always been disappointed that so much of it doesn’t get used,” he said. Humans& aims to create AI systems that can genuinely understand users’ goals, values, and emotions, enabling more effective and trustworthy human-AI collaboration. Zelikman envisions these models playing a transformative role in solving complex societal challenges, such as advancing medical research and accelerating breakthroughs in areas like cancer treatment. His approach stands in contrast to the current wave of AI development, where many companies prioritize scale and speed over depth of understanding. The surge in high-valuation seed rounds—like Thinking Machines Labs, founded by former OpenAI CTO Mira Murati, which raised $2 billion at a $12 billion valuation—reflects venture capital’s bet on small, elite teams to deliver the next major AI leap. While Zelikman has not publicly confirmed the funding details, his vision for emotionally intelligent AI has already drawn significant interest. The success of Humans& will depend on its ability to deliver on the promise of more human-centered AI—where the goal isn’t just to answer questions, but to truly understand and support people.