Home / Technology / AI Ambition: Unpacking the New Foundation Model Scale
AI Ambition: Unpacking the New Foundation Model Scale
24 Jan
Summary
- A new five-level scale measures AI labs' ambition, not just profit.
- Humans& aims to redefine workplace tools but remains vague on specifics.
- Safe Superintelligence's focus is scientific, insulated from commercial pressures.
A new framework assesses AI foundation model companies on a five-level scale, measuring ambition over immediate profits. This scale helps clarify the diverse goals of emerging AI labs, distinguishing between those pursuing massive commercial success and those focused on research.
Prominent labs like OpenAI, Anthropic, and Gemini operate at Level 5, already generating substantial revenue. Newer ventures present a more complex picture. Humans&, for instance, is placed at Level 3, aiming to revolutionize workplace software, though concrete product details are scarce.
Thinking Machines Lab faces scrutiny due to executive departures, raising questions about its initial Level 4 ambition potentially being closer to Level 2 or 3. World Labs, led by Fei-Fei Li, has demonstrated rapid progress, shipping a world-generating model and a commercial product, suggesting a Level 4 trajectory.
Safe Superintelligence (SSI), founded by Ilya Sutskever, is positioned at Level 1, prioritizing scientific advancement and commercial insulation. However, SSI could pivot towards commercialization if research timelines extend or if its powerful AI impacts the world significantly.
