Richard Socher's new startup aims to build artificial intelligence systems capable of researching and improving themselves indefinitely. The founder says the company will deliver actual products, not just research.
Richard Socher, a prominent AI researcher, has launched a new venture backed by $650 million in funding to pursue a bold goal: creating AI systems that can autonomously research and enhance their own capabilities.
The startup represents a significant bet on recursive self-improvement—the concept that sufficiently advanced AI could optimize itself without human intervention. Socher has committed to shipping commercial products alongside this research, distinguishing the effort from purely academic endeavors.
Self-improving AI touches on foundational questions in the field. Current large language models and AI systems require human feedback and updates. A system capable of independent improvement could theoretically accelerate development cycles and reduce reliance on manual optimization.
However, the technical challenges are substantial. Self-improvement requires systems to identify their own limitations, propose solutions, and validate improvements—all while maintaining safety and reliability. The computational costs and potential risks of uncontrolled optimization remain open questions.
Socher's background includes leadership roles at Salesforce Research and founding Roblox's AI division. His track record suggests serious technical capability, though the $650 million budget indicates the challenge ahead.
The startup enters a crowded landscape where major tech companies and well-funded labs are pursuing similar research. OpenAI, DeepMind, and others have published work on AI self-improvement mechanisms, though practical implementations remain limited.
The commitment to shipping products sets this venture apart from pure research initiatives. This approach creates pressure to demonstrate near-term commercial value while pursuing longer-term self-improvement capabilities.
Success would reshape AI development timelines and economics. Failure would validate skepticism about near-term recursive self-improvement. The startup will operate in this space between ambitious technical goals and market expectations.
Uber's weekly AI agent requests have grown nearly tenfold since February, yet the company has held spending flat since April after exhausting its entire 2026 AI budget in Q1.
A recent paper shows artificial intelligence often diagnoses and treats patients better than human physicians. The findings are prompting difficult conversations within the medical community about the profession's evolving role.
The Relay Q, launching next year, represents the latest push to establish voice as the primary interface for human-computer interaction, challenging the keyboard's decades-long dominance.
An Anthropic researcher demonstrated automated systems that can identify and correct misaligned behaviors without compromising overall performance. The systems improved on all 10 tested benchmarks measuring specific problematic outputs.