AI21 Labs, a privately owned American company founded in 2017, develops and provides access to language models. Employing approximately 230 people, the company's core business is software development focused on creating tools that enable text generation and comprehension capabilities.
AI21 Labs explored the current state of AI, contrasting reality with hype, focusing on large language models and the neurosymbolic future. Challenges in transitioning AI from pilot programs to full production within enterprises were also examined, highlighting economic reliability and control issues hindering adoption. The future of human-AI collaboration was discussed, including concepts like cyborg organizations and centaur benchmarks. The limitations of transformer architecture in large language models were noted, specifically the quadratic complexity of the self-attention mechanism and its impact on computational expense and memory usage. Concerns regarding data management and confidence in AI systems were raised, particularly in sectors deploying AI agents and integrating AI-driven processes where data spans extensive and complex knowledge bases. Finally, the challenges of implementing enterprise AI in the retail sector were addressed, emphasizing the need for investment in data quality, human oversight, and cross-functional alignment for successful transitions to advanced AI ecosystems. The importance of AI governance, especially with the rise of private AI, was also highlighted.
Subscribe for full access to AI21 Labs's products in full detail 
