During Google I/O 2023, CEO Demis Hassabis declared that we stand at the foothills of AI’s scientific singularity. Yet, his lofty rhetoric about autonomous scientists contrasts sharply with real-world tools like WeatherNext, which saved lives during Hurricane Melissa.
The key now is whether agentic systems—like OpenAI’s models proving mathematical conjectures—can overtake specialized AI tools in scientific research. Google's Gemini for Science suggests they're eyeing this future, but it’s unclear if the public and scientists will embrace such a paradigm shift.
For now, AlphaFold and similar tools remain indispensable. But as coding capabilities improve, we might see fewer specialised systems and more agentic agents driving scientific progress. This could spell an intriguing new era where AI collaborates—or competes—with humanity in the laboratory.
The question looms: will machines become our partners or surpass us? Google’s move hints at a path towards the latter, but only time will tell if this vision prevails.
