Patrick Schoo, a professor at UC Berkeley
OpenAI Deep Research is probably my second experience of feeling the presence of artificial intelligence generalization (AGI).
- This tool greatly expands my thinking ability. It synthesizes knowledge conceptually and meaningfully, meta-analyzes and summarizes data, and organizes my disorganized scientific intuition in my head. It also reinforces my reasoning with solid citations, connects concepts between fields, and analyzes raw data in multiple modal ways.
- Research is both science and art. Skilled researchers are differentiated by their high level of perspective and ability to raise and explore problems. It’s like ‘Prompt Engineering to Nature’. I predict that the ability of Deep Research to bring out its true potential will make a big difference even among researchers, and those who are good at it will be far ahead of the average user.
- The education system needs to change. Universities should offer lectures that teach prompt engineering in each field. This will be the most valuable class in the future.
- Powerful technology frontier evolves according to the level of abstraction we can work with (conceptually, we could call it Abstraction Scale). Just as we have evolved from assembly to C, Python, and agent-based programming, research meta-processes are increasingly moving towards automation and recursive execution. This offers interesting implications for investment and business operations.
- Current laboratories in biology are no different from those of the 1990s or even the 1960s. However, change is now beginning. Humanoid robots, guided by physically intelligent and autonomous AI models, will conduct 24/7 research in future laboratories. Our assumptions will change, and it is possible that v1 robot deacons will be available to American households as early as the end of 2025 or early 2026.
- The best interdisciplinary research essentially comes from bilingual or multilingual scientists. This is because they have excellent abilities to intuitively capture meaningful problems across different fields (e.g., chemistry + biology = biochemistry or chemical biology, physics + cell biology = molecular biology, economics + mathematics + business administration = financial engineering, etc.) In the agent era, high-quality ideas and conceptual connections will be abundant, but the most important limiting factor will be the advanced technology to actually implement these ideas.