How Do We Build Labs that Will Learn?

by John F. Conway, Founder & CVO, 20/15 Visioneers

For decades, laboratory transformation was largely about moving from paper to digital. We built ELNs, LIMS, instrument integrations and scientific data platforms to capture what scientists were doing and make that information searchable, connected and compliant.

But what happens when the objective is no longer simply to capture science digitally — but to make the laboratory itself dramatically more intelligent?

The next evolution of the laboratory is emerging at the intersection of human intelligence, artificial intelligence, automation and increasingly sophisticated experimentation. The ambition is not simply to put an AI assistant beside every scientist. It is to create a scientific environment in which models can help formulate hypotheses, design experiments, interpret results, and continuously learn from what happens in the physical laboratory.

This raises a fundamental question: What would a super-intelligent laboratory actually look like?

The answer may be less about replacing human intelligence and more about amplifying it. Scientists still provide domain expertise, intuition, creativity, judgment and the ability to recognize when something unexpected matters. AI and automation can provide scale, speed, pattern recognition and the ability to explore experimental spaces far beyond what humans can practically investigate themselves.

But there is a catch: intelligence is only as good as the evidence it learns from.

The laboratory of the future therefore requires much more than better AI models. It requires better scientific data — structured, contextualized, traceable, model-quality data generated through carefully designed experiments. It requires instruments that can communicate, automation that can execute, informatics that can contextualize, and experimental systems capable of feeding results back into the models.

Think about what happened with the latest generation of telescopes and advanced microscopes. We did not simply build better ways to look at the same things. We built instruments capable of revealing phenomena we could not previously see. Suddenly, the challenge was no longer just observation — it was understanding the enormous volume and complexity of what we could now observe.

The same transformation is beginning to happen in the laboratory.

Advanced experimentation, high-throughput automation, multimodal data, simulation, AI-driven experimental design and increasingly autonomous workflows are giving us the ability to see, test and explore more of scientific space than ever before.

So perhaps the journey from the Paperless Lab to the Smart Lab is only the beginning.

The next destination may be the Superintelligent Lab — not a laboratory without humans, but a laboratory where humans and machines continuously amplify one another, where every experiment improves the next decision, and where the laboratory becomes a learning system.

The question is no longer simply:

“How do we digitize the laboratory?”

It is:

“How do we build a laboratory that can learn?"