Scientists can grow three-dimensional neural tissue, stimulate it through electrode arrays, record its activity, and use changes in that activity as computational output. Prior experiments have demonstrated simple pattern and prediction tasks, while a wider research community is building common methods across neuroscience, tissue engineering, electronics, and machine learning. The field now has dedicated funding and a growing body of reviews. Its credible present is a controllable biological research system, not a conscious desktop computer or a general-purpose AI.

Current organoids are small, variable, difficult to maintain, and far from the structure or capability of a human brain. Researchers have not shown that living systems outperform silicon on useful workloads once culture, interfaces, energy, equipment, and labor are counted. Claims about learning can rest on narrow definitions and tiny tasks. Moral status is uncertain, but uncertainty does not justify sensational claims of awareness. Donors may also have agreed to medical research without anticipating that their cells would become part of a computing experiment.

Watch for repeatable benchmarks across laboratories, stable cultures over long periods, clear comparisons with silicon and neuromorphic chips, and evidence that a biological system adds scientific value. Governance should specify cell provenance, donor permission, allowed uses, monitoring, and stopping rules before capability rises. Researchers also need shared language that distinguishes adaptation, information processing, sentience, and consciousness. The best near-term projects may use the systems to study learning or neurological disease rather than claim a new class of commercial computer.