Thinking Machines: Mira Murati’s new AI startup | A compact guide to large language models pdf free | Chatgpt as a large language model i cannot | Llm ai | Turtles AI

Mira Murati, former CTO of OpenAI, launches Thinking Machines Lab, a startup focused on collaborative and customizable AI. The company aims to overcome the current limitations of AI technology through multimodal models, advanced infrastructure, and a strong commitment to security and transparency.

Key Points:

• New AI startup: Mira Murati founds Thinking Machines Lab with ex-OpenAI and DeepMind experts.
• Customizable AI: Systems designed to adapt to specific user needs.
• Human-machine collaboration: Focus on multimodal models for more natural interactions.
• Security and transparency: Open research, data sharing, and abuse prevention.

Former OpenAI CTO Mira Murati has officially announced the launch of Thinking Machines Lab, a new startup dedicated to developing AI systems that can adapt to individual users and integrate more effectively with human expertise. The company, which includes John Schulman as Chief Scientist and Barret Zoph as CTO, aims to fill gaps in the AI ​​sector, in particular by improving the understandability, personalization, and reliability of advanced models.

In the current landscape, technological progress has led to a rapid development of AI capabilities, but at the same time has highlighted a significant discrepancy between the speed of innovation and the scientific understanding of these systems. Knowledge of how these models are trained remains confined to major research laboratories, limiting the possibility of an informed public debate and reducing the accessibility of AI tools for more effective and targeted use. Thinking Machines Lab was born to directly address these issues, proposing a more transparent and collaborative approach to research and development in the field of AI.

One of the pillars of the startup is the creation of multimodal models, capable of interacting with humans in a more intuitive and adaptable way. The goal is to build systems that are not simply automation tools, but real collaborative partners in different professional fields. While current AIs excel in areas such as programming and mathematics, Thinking Machines Lab wants to go further, creating solutions capable of covering the entire spectrum of human skills and enabling applications in new contexts.

Fundamental to this vision is the quality of the technological infrastructure, an aspect on which the company places great emphasis. The efficiency, reliability and scalability of research and development systems are considered absolute priorities, avoiding shortcuts that could compromise safety or usability in the long term. Furthermore, multimodality will be a key element, improving the ability of machines to understand diversified inputs and to operate in more complex and multifaceted contexts.

The company’s approach to AI safety is based on an empirical and iterative model, combining advanced research and practical testing. Strategies include preventing the misuse of models, sharing best practices with the entire scientific community, and supporting external research through the disclosure of code, datasets, and technical specifications. Thinking Machines Lab also promotes the regular publication of studies, technical papers, and open-source projects, believing that AI progress must be a shared and collective process.

Mira Murati, who led the development of flagship products such as ChatGPT, DALL-E, and Codex during her experience at OpenAI, brings with her a clear and ambitious vision: to build more capable, adaptable, and secure AI systems, rethinking not only the evaluation metrics but also the very goals of AI. With a team that includes former members of OpenAI, Google DeepMind, and Character AI, the startup is actively recruiting experts in machine learning and advanced research to accelerate its development.

In the context of an ever-evolving sector, with emerging startups such as Anthropic and Safe Superintelligence founded by former OpenAI executives, Thinking Machines Lab comes in with a distinctive proposition, aiming to make AI more accessible, collaborative and capable of generating concrete value in the real world.