tl;dr Until we understand how complex and chaotic systems^[1] in nature work, "natural is better" is a valuable heuristic. It applies wherever science does not understand things well enough, especially when the stakes are high.

If you throw a stick into a forest it will biodegrade quickly and not affect the forest much; this is because forests are used to sticks.^[2] Throw a piece of plastic in the forest and you will affect the forest a lot more. Not because plastic is inherently bad; but because the forest has not evolved to deal with plastic and will thus have a harder time breaking it down. There's nothing inherently unhealthy or destructive about plastic, in the same way that there is nothing inherently unhealthy or destructive about sticks: natural systems in the forest are just not used to coping with the former.

Some other things natural biological systems tend to not be used to: nuclear radiation, the vacuum of space, some pesticides, some GMOs, some fertilizers, a global rising of temperatures, asteroid impacts, supernovae, acidification of the oceans, sudden changes in atmospheric composition, etc. The status quo of life on Earth has faced all these things at various moments throughout history. They are examples of perturbations in a very complex system, but, like plastic, they are not inherently bad; they are just things biology has to adapt to.

I feel like this is a fact that both "natural is always better!" types and "I'll roll my eyes at that" types forget. Here's how I see it:

Science is a cool new trick the monkeys figured out which lets them make maps of reality that are so accurate that they know where the levers are, and can pull them in a way that makes the universe behave a certain way. Structure atoms right and the universe will, among many other things, do math for you, generate photons, detect an atom, establish probabilities, store information, land you on the moon, change states of matter, ...

Sometimes, however, science doesn't know where all the levers are: sometimes systems, like forests, are so complex and chaotic that pressing one lever might set off a cascade and bring about a whole slew of unintended consequences. In principle, give science enough time and it will figure out how to press the right levers safely. But science is a rather new thing and the amount of unknowns in nature is vast.

Science has the potential to be much better than nature. We know what levers to pull for a person to be immune to malaria, and pulling those levers on a human puts them in a healthier state than they would be naturally. At the end of the day, understanding the universe well enough not to mess it up when you pull its levers makes the artificial better and healthier than the natural. Science triumphs over nature in the long run. But in the meantime, "natural is better" is often a valuable heuristic.

Earth is under siege: science is still blind to most things and the road is long to find all the levers. Here are a few excessively complicated natural systems scientists are trying to figure out right now: cancer (biology), alignment (programming), particle behavior (quantum dynamics), protein folding (chemistry), aging (biology), the brain (neuroscience), the mind (psychology), the Riemann hypothesis (math), the Fermi Paradox (a whole slew of disciplines), ...

"Natural is better" as a heuristic

Medical trials can take decades because there is a lot at stake in medicine; namely, the health of a human being. Similarly, there are other natural systems that we do not understand and that we should be very careful with. Making the universe do the type of math which runs artificial general intelligence, for example, sounds like something with a lot of interconnected levers that we should understand perfectly before pulling. The universe is a ridiculously dangerous booby trap and we should avoid rash moves.

When the stakes are high and you don't understand how a forest works, it's safer to throw a stick in than it is to throw plastic. This is why I sympathise with ludites and other cave-dwellers: it's very easy to catch a glimpse at what we don't know and come to the conclusion that frankly we're just monkeys who, through a strange cascade of curious circumstances, found themselves knee-deep in gasoline with nothing but Prometheus matches in our hands.^[3] Toby Ord was right when he described our current position as being a precipice, and I don't blame the people who want us to turn back on our tracks.

Where this heuristic applies in practice

We know how to make safe nuclear energy. It takes decades of careful research, but we also know how to make safe medicine and GMOs. Where would this heuristic actually apply? The heuristic mostly works, I think, in how we structure our lives. There's a general consensus on LessWrong and elsewhere that a whole plethora of problems are caused today, mostly due to an overabundance of superstimuli. As much as the line might be overused, technology has supplied us with a whole lot of "cookies": things that are unhealthy but that we consume anyway because they are specifically designed to cater to our evolution-derived inputs. "Natural is better" would mean structuring your life, at least in part, based on how your body is designed to live its life; don't fragment your mind by paying attention to byte-sized content; don't get riled up in the carefully designed polarized world of "reality TV politics"^[4]; don't eat unhealthily or avoid sports. In a way the entire mission of rationality is just that: patching the parts of our mind that are not well-suited for the modern world. I don't know how helpful it would be for you to read up on how our ancestors lived, but there sure is a lot of content on LessWrong to explain where your mind is misguiding you for the only reason that the environment it had adapted to has drastically changed. How similar is your life to that of a homo sapiens from 12,000 years ago? If you made it more similar, would that help you? This is a case where "natural is better" makes sense.

Thanks to Matthew Barnett for pointing out that I hadn't supplied a useful-enough example of this heuristic in practice.