A chemical engineer at Carnegie Mellon University, Gabriel Gomes, is pioneering a shift in how scientific experiments are conducted. His work centers on automating laboratory research using artificial intelligence and robotics, aiming to accelerate discovery while reducing human error and laborious processes. The core of this effort is Coscientist, an AI agent built upon large language models like GPT-4, designed to bridge the gap between complex code and intuitive scientific inquiry.
Gomes’s journey began in rural Brazil, where access to technology was limited. He became the first in his family to attend university, driven by a professor’s challenge to “solve” chemistry through computation. His expertise eventually led him to advise the White House on the potential of intelligent systems. His work now addresses a fundamental challenge in modern science: making powerful new tools accessible to researchers who may lack extensive coding skills.
The Problem With Current Labs
Many modern labs are being built with automated systems, but these rely on programming knowledge that many chemists and biologists don’t have. This creates a barrier to entry: scientists may not use state-of-the-art facilities if they cannot easily interface with the equipment. Gomes recognized this issue and aimed to create a solution where researchers could interact with automated labs using natural language instead of code.
How Coscientist Works
Coscientist operates like an intelligent assistant for lab experiments. Researchers can instruct the AI in plain language — for example, “Draw something cute on the target plate,” and the system will translate that into robotic actions. In one early test, the system successfully drew a fish on a chemical plate without explicit programming. This illustrates the potential to automate tasks that typically require detailed robotic instruction.
The AI also helps collect experimental data at scale. By automating repetitive processes such as measuring reaction kinetics, Coscientist allows researchers to generate datasets that would otherwise be impractical to acquire manually. This opens up new avenues for data-driven discovery in areas previously limited by human constraints.
Impact on Research
The introduction of Coscientist has already transformed the workflow in Gomes’s research group. New students with limited programming backgrounds have quickly adapted to using the system, accelerating their contributions to complex projects. The AI lowers the barrier to entry, enabling faster learning and more efficient experimentation.
The long-term implications are significant. By removing human bias and reducing labor requirements, Coscientist could unlock scientific areas previously considered too tedious or impractical to explore. This includes high-throughput screening of chemical reactions and the creation of large-scale datasets for machine learning applications.
Cautionary Notes
While AI offers tremendous potential, Gomes cautions against uncritical reliance on large language models. Researchers must still validate results and ensure the AI’s instructions align with scientific principles. Despite the rapid advancements, human oversight remains crucial for maintaining accuracy and preventing errors.
The rise of AI-powered labs is poised to reshape scientific research. By automating tedious tasks and democratizing access to advanced tools, systems like Coscientist could accelerate discovery and unlock new frontiers in chemistry and beyond.
