To me, betting on the AI future is delusional optimism (or something akin to religious faith). I think slowing down “progress” is better than speeding it up, since all these problems are caused by “progress” in the first place.
The existence of LLMs has only made society worse so far, and I don’t expect taking a bad idea further will turn it into a good idea. But while I do not agree with your proposed solution, I do agree with the problem
I don’t think these problems were caused by progress since you saw similar things in Rome. > The existence of LLMs has only made society worse so far, This is just not true. Concretely how have LLMs only made society worse so far?
Fair, main empires have collapsed, so the cause cannot be purely modern. But I can see a lot of similarities in how they collapse. Here’s a quote from “The Fate of Empires and Search for Survival” (1978), about the Arab Empire on the tenth century: “Many women practised law, while others obtained posts as university professors. There was an agitation for the appointment of female judges, which, however, does not appear to have succeeded. Soon after this period, government and public order collapsed, and foreign invaders overran the country. The resulting increase in confusion and violence made it unsafe for women to move unescorted in the streets, with the result that this feminist movement collapsed”. Similar issues happened in rome: “The later Romans complained that, although Rome ruled the world, women ruled Rome”
It also mentions sexual degeneracy, corruption, mass-immigration into capital cities, and the decline of religion. Sexual freedom, the decline of religion, and immigration are all considered “progress” today. These topics are all controversial now, but the author doesn’t strike me as an ideologue.
Concretely how have LLMs only made society worse so far?
Let me first state that I can think of no benefits except for convenient access to common information. Downsides of LLMs include harming education (Students cheat with LLMs, teachers might also use LLMs). Harming employment (automated reading and writing of CVs). A massive increase in slop (LLMs makes mass production of low-quality content trivial, and those who use LLMs are rarely capable of producing high-quality content). And a general downgrade in the quality of fields where AI integration has been attempted (e.g. The problems with recent Windows updates). I haven’t even started on AI psychosis and the price of ram sticks.
They’ll be used for automated moderation (terrible idea, terrible results) and various things which further erode online privacy (perhaps selling peoples private information as a way to make up for their lack of profitability). I strongly believe that shareholders have been mislead about the capabilities of LLMs and that everyone else suffers as a result. Would you agree if I said that the value of LLMs lie in their future potential rather than their current abilities?
I know many people working in tech companies who already use frontier LLMs to do tasks that interns would previously do. Most of the economic value of course still lies in the future. I wonder if there is a canonical source explaining these things. Perhaps the METR plot?
To me, betting on the AI future is delusional optimism (or something akin to religious faith). I think slowing down “progress” is better than speeding it up, since all these problems are caused by “progress” in the first place.
The existence of LLMs has only made society worse so far, and I don’t expect taking a bad idea further will turn it into a good idea. But while I do not agree with your proposed solution, I do agree with the problem
I don’t think these problems were caused by progress since you saw similar things in Rome.
> The existence of LLMs has only made society worse so far,
This is just not true. Concretely how have LLMs only made society worse so far?
Fair, main empires have collapsed, so the cause cannot be purely modern. But I can see a lot of similarities in how they collapse. Here’s a quote from “The Fate of Empires and Search for Survival” (1978), about the Arab Empire on the tenth century: “Many women practised law, while others obtained posts as university professors. There was an agitation for the appointment of female judges, which, however, does not appear to have succeeded. Soon after this period, government and public order collapsed, and foreign invaders overran the country. The resulting increase in confusion and violence made it unsafe for women to move unescorted in the streets, with the result that this feminist movement collapsed”.
Similar issues happened in rome: “The later Romans complained that, although Rome ruled the world, women ruled Rome”
It also mentions sexual degeneracy, corruption, mass-immigration into capital cities, and the decline of religion. Sexual freedom, the decline of religion, and immigration are all considered “progress” today. These topics are all controversial now, but the author doesn’t strike me as an ideologue.
Let me first state that I can think of no benefits except for convenient access to common information. Downsides of LLMs include harming education (Students cheat with LLMs, teachers might also use LLMs). Harming employment (automated reading and writing of CVs). A massive increase in slop (LLMs makes mass production of low-quality content trivial, and those who use LLMs are rarely capable of producing high-quality content). And a general downgrade in the quality of fields where AI integration has been attempted (e.g. The problems with recent Windows updates). I haven’t even started on AI psychosis and the price of ram sticks.
They’ll be used for automated moderation (terrible idea, terrible results) and various things which further erode online privacy (perhaps selling peoples private information as a way to make up for their lack of profitability). I strongly believe that shareholders have been mislead about the capabilities of LLMs and that everyone else suffers as a result. Would you agree if I said that the value of LLMs lie in their future potential rather than their current abilities?
I know many people working in tech companies who already use frontier LLMs to do tasks that interns would previously do. Most of the economic value of course still lies in the future. I wonder if there is a canonical source explaining these things. Perhaps the METR plot?