Nature Biotechnology published a review outlining a vision for generalist biological AI—models that can perform diverse tasks across biological domains. The paper evaluates promises and pitfalls, including cross‑task transfer, reproducibility, data biases and the need for standardized benchmarks and datasets. Authors argue that properly validated generalist models could accelerate target discovery, experimental design and interpretation, but stress rigorous evaluation and clear reporting to avoid overclaiming capabilities. The review also highlights governance, data accessibility and integration challenges as adoption scales across academia and industry.