Reinforcement learning may boost LLMs today, but it cannot deliver safe, long-term intelligence. We argue for System 2 learning instead.

Elicit develops LLM-based auto-evals to balance scale, trust, and flexibility, ensuring reliable scientific reasoning at superhuman speed.

Not every AI role needs a PhD in ML. A good AI engineer can bridge full-stack skills with LLM expertise, making AI practical in real products.

Elicit implements SPLADE to leverage semantic search while retaining consistent rigor and accuracy.

Elicit's nuanced stance on coding assistants in interviews: welcome as tools for testing insight, but not as replacements for genuine reasoning.

Don’t confuse embeddings with memory. To build effective RAG, start with a search engine, not just a vector database.