Artificial intelligence has long supported scientific research by analyzing data, reviewing literature, and assisting with simulations. Researchers at IIT Delhi have now taken a decisive step further. Their newly developed AI agent, AILA (AI Laboratory Assistant), does not merely assist scientists—it actively performs real-world laboratory experiments.
This development marks a shift from AI as a digital helper to AI as an experimental collaborator.
What Is AILA?
AILA is an AI agent designed to autonomously conduct physical scientific experiments. Unlike conventional large language models that generate answers based on existing data, AILA is directly connected to laboratory equipment. It can:
Interpret natural-language prompts
Write and execute instrument-control code
Operate complex lab machinery
Collect experimental data
Analyze results and present conclusions
In its current implementation, AILA is integrated with an Atomic Force Microscope (AFM), a highly precise instrument used for nanoscale material characterization.
Moving Beyond Text-Based AI
Traditional AI tools such as chat-based models are limited to information already available in datasets or on the internet. AILA breaks this limitation by generating new scientific data.
Researchers can issue commands such as:
“Capture a 10 nm × 10 nm AFM image and calculate surface roughness.”
AILA then executes the entire workflow—approaching the sample, tuning microscope parameters, scanning, storing the image, and analyzing it—without manual intervention.
This ability to close the loop between instruction, execution, and analysis is what sets AILA apart.
Why Atomic Force Microscopy?
The research team selected AFM as a proof of concept because it is:
Widely used across materials science, electronics, and biology
Technically complex, requiring careful parameter tuning
Representative of many advanced laboratory instruments
By successfully automating AFM workflows, the team demonstrated that their framework can, in principle, be adapted to other experimental platforms.
Impact on Research Productivity
One of the most significant outcomes of AILA is time efficiency. Tasks that previously took an entire day—such as parameter optimization and repeated trial-and-error experiments—can now be completed in minutes.
Key advantages include:
24/7 operation without fatigue
High-throughput experimentation
Consistent execution and logging
Reduced manual workload for researchers
Scientists remain responsible for defining goals and interpreting results, while AILA handles repetitive and time-consuming tasks.
Accuracy and Reliability
To evaluate reliability, the researchers tested multiple language models across 100 standardized experiments. The best-performing models achieved close to 80% accuracy in executing tasks correctly.
Importantly, AILA maintains detailed logs of every step it performs. This transparency allows researchers to:
Review decision-making processes
Validate experimental outcomes
Identify and correct errors
Human oversight remains central, particularly for safety-critical operations.
Applications Beyond the Lab
Although currently demonstrated with AFM, the broader framework behind AILA has implications across multiple domains:
Energy: Discovering improved battery and superconductor materials
Environment: Developing and testing advanced pollution sensors
Healthcare: Accelerating biomaterials and diagnostic research
Aerospace: Evaluating materials under extreme thermal conditions
In each case, the common bottleneck is experimental throughput—precisely where AILA provides value.
Human–AI Collaboration, Not Replacement
The researchers emphasize that AILA is not designed to replace scientists. Instead, it acts as a co-pilot:
Humans define hypotheses and research directions
AILA executes experiments rapidly and consistently
Humans interpret results and make final decisions
This collaboration allows scientists to focus on creativity, theory, and innovation rather than routine instrument operation.
A Step Toward the Future of Science
AILA represents a meaningful transition in how research is conducted. By bridging AI reasoning with physical experimentation, IIT Delhi has demonstrated a model for the future laboratory—one where AI systems actively participate in discovery.
As these systems mature and expand to other instruments, autonomous experimentation could become a standard component of scientific research, reshaping productivity, training, and innovation worldwide.