Re: AI is not “just a tool”

Ah, moral absolutism.

In recent years, I've noticed people making it a point to be more aware of how they're spending money. Some people boycott Target because they canceled their DEI policies. Some people boycott Nike because they supported trans athletes. The Home Depot co-founder was problematic, so we're not supposed to shop there anymore. We shouldn't buy Chick-fil-A because they do not support non-traditional marriage. Most of the anti-AI positions I read are in the same spirit as these.

If a person pulls on that thread too long, then they must back themselves into such a corner that they can't morally use any technology or even be a member of any country. What was the cost in terms of human life and environmental resources of the industrial revolution? What were the motivations behind the people polluting the world and taking advantage of the poor?

How can anyone morally live in the United States? Europeans colonized it. They killed the natives and drove them from their homes, their lands taken from them by force.

What virtues is a person signaling by remaining a member of such societies and countries?

LLMs had a messy and questionable start, but there is real capability there and it has changed how the world does things with computers. I see a path where I can buy a sufficiently capable computer for my homelab, then run my own model to empower myself with more digital freedom and sovereignty.

We should sufficiently regulate the companies building the frontier models, just like any other company, tech or otherwise, so they align with the values of the society they operate in. I think Amazon buying rare books, which it then destroys as part of the scanning process for LLM model training, is morally reprehensible. I think the LLM companies scraping the entire internet to train their models without recompense to the people they took the information from is obscene. But the scanned books are already destroyed and gone. The tech companies have already taken, stored and used the data. The genie is out and there's no putting it back.

Using an open model is not morally equivalent to destroying rare books for AI training, stealing everyone's content on the internet, or companies trying to replace employees with AI chatbots. It's just not. The problem with this type of thinking is that it's guilt by association, once, twice and sometimes thrice removed.

Look, we should all do what we feel is right. To this day, I will not spend money at Chick-fil-A because I don't want my money going towards homophobic advocacy. But I think this idea that people can just opt out of all AI is silly. Somewhere in your supply chain, there is someone using AI to build something you like and use.