Thinking Machines Lab AI News & Updates
Thinking Machines Lab Develops Method to Make AI Models Generate Reproducible Responses
Mira Murati's Thinking Machines Lab published research addressing the non-deterministic nature of AI models, proposing a solution to make responses more consistent and reproducible. The approach involves controlling GPU kernel orchestration during inference processing to eliminate randomness in AI outputs. The lab suggests this could improve reinforcement learning training and plans to customize AI models for businesses while committing to open research practices.
Skynet Chance (-0.08%): Making AI models more deterministic and predictable reduces one source of unpredictability that could contribute to AI safety risks. More consistent AI behavior makes systems easier to control and understand, slightly reducing alignment concerns.
Skynet Date (+0 days): While this improves AI reliability, it doesn't fundamentally accelerate or decelerate the timeline toward potential AI control problems. The research addresses technical consistency rather than capability advancement that would change risk timelines.
AGI Progress (+0.03%): Improved determinism and enhanced reinforcement learning efficiency represent meaningful technical progress toward more reliable AI systems. Better RL training could accelerate development of more capable and controllable AI models.
AGI Date (+0 days): More efficient reinforcement learning training and reproducible responses could modestly accelerate AGI development by making AI training processes more reliable and effective. However, this addresses training efficiency rather than fundamental capability breakthroughs.
Former OpenAI Leadership Joins Mira Murati's AI Startup as Advisers
Thinking Machines Lab, the AI startup founded by former OpenAI CTO Mira Murati, has added two prominent ex-OpenAI leaders as advisers: Bob McGrew, former chief research officer, and Alec Radford, a pioneering researcher behind GPT technology. While the startup's specific research agenda remains vague, it aims to build AI systems that are "more widely understood, customizable, and generally capable" than current options.
Skynet Chance (+0.04%): The concentration of top AI talent from OpenAI in a new venture increases competitive pressure in advanced AI development, potentially accelerating capability advances while diluting established safety cultures, though the emphasis on making AI "more widely understood" suggests some focus on transparency.
Skynet Date (-1 days): The creation of a well-funded competitor with elite talent from OpenAI intensifies the competitive landscape for advanced AI development, likely accelerating timeframes as multiple groups pursue similar cutting-edge capabilities in parallel.
AGI Progress (+0.03%): The migration of key talent responsible for OpenAI's most transformative technologies to a new venture focused on "generally capable" AI systems represents a moderate redistribution of expertise rather than new capabilities, though it may lead to novel approaches through competitive pressure.
AGI Date (-1 days): The formation of an additional well-resourced lab led by the architects of breakthrough AI systems like GPT intensifies competition in advanced AI development, likely accelerating progress toward AGI through parallel efforts and competitive dynamics.
Former OpenAI Leaders Launch Thinking Machines Lab to Build More Customizable AI
Former OpenAI CTO Mira Murati has launched Thinking Machines Lab, a startup focused on developing more customizable and capable AI systems that address key gaps in current AI technologies. The company, which includes OpenAI co-founder John Schulman and other high-profile AI researchers, aims to build frontier multimodal systems for applications in science and programming while emphasizing AI safety.
Skynet Chance (+0.01%): The emphasis on building highly capable frontier models increases potential risks, but the explicit focus on customizability, safety practices, and sharing alignment knowledge provides some counterbalance. Their stated commitment to understanding systems indicates awareness of control issues.
Skynet Date (-1 days): The formation of another highly-credentialed team pursuing frontier capabilities at the limits of AI, particularly with multimodal systems for science and programming, will likely accelerate development timelines toward more advanced systems with potentially unpredictable emergent behaviors.
AGI Progress (+0.03%): The assembly of key technical leaders from OpenAI, including those who helped develop ChatGPT and other breakthrough systems, focusing explicitly on frontier multimodal models represents a significant concentration of talent that will likely drive substantial technical progress toward more AGI-like capabilities.
AGI Date (-1 days): The emergence of another well-funded company founded by architects of today's most advanced AI systems, explicitly focused on frontier capabilities in domains like science and programming, will likely accelerate development timelines through additional competitive pressure and parallel research efforts.