I disagree with the author on the below:
>With GPT-3, we now have an infinitely-scalable technology that is years away from being able to enrich our lives, but is already more than capable of drowning out all remnants of authentic content on the internet. And because you can leverage this to earn money or sway opinions, that outcome is probably hard to avoid.
I can't speak for GPT-3 specifically as I haven't spent enough time with it, but it's open equivalent Bloom is very useful today. Author says it is far from being able to "enrich our lives". I think otherwise. The whole "AI revolution" definitely did enrich my life. It renewed my passion for programming in general. The possibilities are truly endless. From translation, to content summarisation and search. Then it is only a step away from enriching other people's lives with products that make them more productive, help them in execution of mundane tasks etc.
I think AIs like bloom will in not to distant time allow us to fix the Internet to how it was before all the shitty content generation replaced genuine content in Google for example.
Also, there is a huge potential to improve our understanding of what it means to think and understand complex concepts. For the very first time we have something that comes very close to such understanding. Something that doesn't have feelings, it can be cut, modified, transformed and studied in myriad of different ways.
Also, believe me when I say the next AI revolution will be with models like Bloom that will be able to run on the edge. This is currently impossible due to their size, but it is very likely we'll find methods to optimise them significantly.
Then there is potential application of such models to decision making. I bet there is ongoing research in that field right now.
Addressing the negative side, "ability to generate garbage content that looks good". The ability has already existed for advanced adversaries. Hostile nations can simply employ people to "generate content". Smaller players have lots of other (inferior) tools. Their content doesn't have to fool people most of the time. It just has to fool various algorithms. Even without advanced AI models the search engines and social networks have been loosing badly in this fight. Existence of those tools will improve the bad actors capabilities, but it will also give good actors the chance to innovate with something new.