If Only Earth was a Product
One thing that has fascinated me as generative AI became a constant theme in the news this year is how rarely anyone questions the underlying assumption that this makes sense (or is somehow inevitable) as a focus for huge resources of money & computing power. These interactive AI tools that are currently being blended in to all sorts of daily software for business, school, art, communication, etc. are regularly given attention- and rightly so- for questions around copyright or safety or job losses or artistic impacts. Meanwhile in a year-end podcast looking at how this tech has been developing, Ezra Klein noted how there isn't a public sector version of ChatGPT and bluntly added that "if the private sector thinks it is worth pumping $50 or $100 billion into these companies so they can help you make better enterprise software, it seems weird to imagine that there are not public problems that have an economic value that is equal to that or significantly larger."
I appreciate his subtle sarcasm there as it's clear to any slightly sentient human that climate change is a public problem with such a voluminous scale of cost for addressing it. I also really appreciated this episode of David Roberts' Volts podcast in which he interviewed Priya Donti, an MIT researcher and the director of the non-profit Climate Change AI. They discuss how a lot of AI and machine learning is very task-based and has real value in addressing climate areas where there are large amounts of data. She gave some examples of using energy grid data for forecasting with solar, or for maximizing efficiency with sensors in the heating and cooling of commercial buildings, or with significantly cutting the development time for designing new batteries.
Yet none of these things have the same sci-fi movie drama of 'artificial general intelligence'. Open AI leads with this statement on their home page: "Creating safe AGI that benefits all of humanity". Many researchers (Gary Marcus is one I find interesting) question whether their process of scraping information right off the internet could even really lead to AGI. But the thing I question most is why they assume it would benefit "all of humanity". When climate change is showing its presence every week, I don't think a world with a language model that can write up a draft marketing plan or AI agents that can make appointments for you is really the essential benefit that will make life so much better.
The other piece that isn't really talked about is the cycle of money. Microsoft spent 50 million dollars in 2017 on a 5 year program called AI for Earth focused on supporting sustainability projects worldwide, according to their website. This sounds great- but that amount shrinks next to the eye-popping number of how much Microsoft invested in OpenAI last year: 10 billion. I saw a news release from 2021 when OpenAI and Microsoft set up a 100 million dollar start-up fund that would focus on supporting startups in "healthcare, climate change, education and AI applications". Yet, this article from this year listed the 12 companies that actually benefited from that money and none of them are climate-related (or health or education ones either). Climate change is so dire already yet it can still be treated like a pet project or just casually removed from the list of areas for the 'captains of capital' to focus on.