The debate about plant intelligence is quite interesting. However, from having just a cursory look of a little bit of reading and one class session about the topic, and a much larger background in computing and artificial intelligence, the resemblance between plant intelligence and AI was quite striking. Plants were described as not being human in intelligence, but still being able to make decisions, and learn. Well, a computer could do that as well! Just as a plant may track towards the sun or learn to not be afraid when it was dropped, a machine learning model can learn to give better outputs, or even in an applied robotics sense move its' limbs in such a way until it figures out how to walk. Just like how plants evolve over time, machines do as well, with both either producing new offspring with slightly more optimal genetics or running through algorithms to slightly fine tune results. There are a few other similarities as well.
The above graph describes how plants act based on processes, sequences of interconnected actions that unfold over time. Quite interestingly, processes is also a computer science term, describing running programs, which are sequences of programs giving instructions that unfold in polynomial time.
Above is an example of a Black Box model Neural Network. A neural network is a type of algorithm that models how neurons work in the brain. The black box model is where the inputs are put in, the outputs are received, and the actual formulas used to transfer input to output is unknown. This means that users can use a neural network without having to understand the high level Mathematics and Computer Science principles behind it. Of course, an expert in the field could look at a model and tell you exactly how the skeleton is made, although as models get more complex and require more and more specified knowledge perhaps a team of experts would be needed. Does any of this sound familiar? With plant behavior, plants receive some sort of input, like sunlight or being dropped, and then give out some output, like moving towards the sun or realizing dropping does not damage it. Are these two processes not the same? Teams of experts are required to decipher why the input turns into the output, but the process still happens the same, no matter how much we know of the black box. Of course, Computer Science is a human made field, and a model has to be coded by a group of people, so there exists the knowledge of why things were programmed the way they were, and the black box would be much clearer than plants. But aren't plants just programmed by God, or evolution? By cracking the natural programming language of plants, it can be deciphered why the inputs are turned into outputs, and the scientists studying plant behavior are trying just that.
So, are plants intelligent? They seem about as intelligent as our most cutting edge machine learning models. Both just take in inputs, either through data or senses, and then run their internal programming to give an output, an action, that takes into account said input data. Of course, this argument could be spread to humans, but the fact that humans are conscious makes that a hard sell to me. Plants, like programs, can do amazing things, and if you consider being able to take in large amounts of input data and giving an appropriate reaction, they are intelligent, but if you consider intelligence to be something more demanding, then they are not. Plants seem to just be natural computers, programmed in an unknown language, being reversed engineered by botanists as we speak.
Comments
Post a Comment