The Book
In his book More Everything Forever, Adam Becker provides a compelling argument, backed by interviews, documented legal proceedings, and other publicly available information, that the current A.I. mania is driven by a cult and the real-world impacts of this cult’s delusions are worse (and more real) than the apocalyptic futures they imagine.
The cult he writes of is not a singular entity but a group of organizations founded by like-minded people. They share similar beliefs and motivations, but you can’t point to a single organization such as Scientology. The groups the author focuses the most on are the Rationalists and the Effective Altruists.
I will try to summarize what these individuals believe…
If we do not create artificial super intelligence (Gen A.I.) and colonize the universe then humanity is doomed. The artificial super intelligence must be developed and trained to work for our benefit at all costs. Human suffering that exists today is outweighed by the future flourishing of human civilization in space, and therefore there should be few to no limits on what we do to achieve it.
Sounds like something out of Asimov, right?
Most of the individuals mentioned in the book are very wealthy, or otherwise have stakes in the A.I. industry. Think people such as Sam Altman, Elon Musk, and formerly wealthy Sam Bankman-Fried. They all share some combination of beliefs which circulate in the Rationalist and Effective Altruist communities. While many of these beliefs sound positive or beneficial on the surface, Becker outlines how they are rooted in the idea that humanity should exist everywhere, forever, and that there should be no limits to our consumption even if it means eating the entire universe.
Many of the ideas generated by these communities are unfathomably nutty sci-fi fever dreams. I will not summarize them here, because you should buy and read the book.
My Thoughts
As a technology person I am used to these hype cycles. Not long ago it was the blockchain. We were all supposed to become knowledgeable of and find ways to use this highly inefficient transaction registry to somehow make everything better, as promised by its hype men and believed by corporate managers and executives. That technology found niches in which it is useful (I assume in fintech, but I don’t work in that realm) but it didn’t change much. The hype bubble burst and made some people a lot of money before it did.
Now it’s all about A.I.. Companies are concerned with keeping up the pace of technology to ensure they stay competitive and relevant. I can’t blame them for ensuring their technology people are aware of these things especially when the A.I. hype men are promising them more “productivity” at lower cost and higher profits. At least I imagine this is what your average corporate boardroom types are thinking. Some of them may believe in the nuttier stuff explained in Becker’s book, but I hope that is not a large percentage.
What we currently refer to as A.I. are actually Large Language Models (LLMs). These are essentially large indexes of words and their relationships to each other as used in language. Companies like OpenAI feed their models truly massive amounts of text scraped from internet sources, books, etc. It is impressive how much data they process, how much smart (human) engineering and labor they expend, and how accurately the LLMs can create sequences of words that make sense in English.
My technical experience with this sort of technology is limited. I have experimented with Natural Language Processing (NLP) and have learned about related concepts. There is some definite utility to this technology to augment human capabilities, but I have not seen anything yet that makes me believe they can replace human capabilities. LLMs can be fantastic accelerators for things such as language translation, parsing and summarizing large amounts of research text and data, or helping with reconstructions of very ancient text. If models are trained for their specific purposes they could be very useful to us indeed, but I do not see much real benefit of the “generalist” models that operate based on a poorly defined context.
I have tried LLM based tools for my work and found the results to be completely underwhelming. The time it takes you to explain a complex problem to a chat bot is much better spent figuring it out with your own brain. You will do a better job, believe in yourself. The LLM based chat bots are nothing more than very adept bullshit generators. There are probably two things I have used one of these things for with any level of success.
- Spitting out a RegEx string
- Sorting a list of lines in a text file by specific criteria
Both of these things are hated, tedious tasks of mine. But the thing is…they aren’t THAT hard and don’t take THAT much time to do myself. It is certainly not worth the massive expenditure of money, electricity, water, computing power, etc. spent on training these models. Every example I have seen of generative A.I. is like this. It’s garbage that pales in comparison to something created by a human being while absorbing resources better spent on human beings.
The utility of this technology compared to its environmental cost is highly questionable. If you ask the Rationalists and Effective Altruists, you may hear something like “Humanity must develop artificial super intelligence and use it to colonize the universe or we will go extinct. The harms caused today are outweighed by the future perpetual survival of humanity.”. Then they may provide astronomically large numbers (of human beings in the future), and remote probabilities to back up their claims. Becker refutes these claims in his book using interviews with actual scientists working on a broad range of related topics. In essence, what the Rationalists and Effective Altruists promise is and perhaps always will be science fiction.
The utility of this technology isn’t my biggest concern. In any field, we try things and see what works. What bothers me is that behind the chatbots and image generators there is a movement led by billionaires who want to be space-imperialists. Instead of pouring their time and money into things that would actually help people here and now, they are working towards an impossible and frankly undesirable sci-fi future where all nature is dominated by tech, and all power is in their hands.
I agree with Becker’s conclusions at the end of the book, wholeheartedly, and I strongly recommend everyone read it.