Is It Rare to Develop New Architectures in Biology?

With a few rare exceptions, most of deep neural networks or other types of AI in biology feels like it's taking something invented somewhere else and carrying it over. To really transfer and develop an algorithm explicitly for the biological problems is pretty rare. And I believe that has to happen in lockstep with then also understanding the peculiarities of said data. But we're not in that regime yet where that's just sitting around on the equivalent of double amp or so.