🌱 Lead Paint Is Amazing

On “Usefulness” and “Harmfulness”

Lenke til norsk versjon

Why are people so down on putting lead in paint? I mean, as Wikipedia puts it:

Lead is added to paint to accelerate drying, increase durability, maintain a fresh appearance, and resist moisture that causes corrosion. (…)

(…) very popular with artists because of its density and opacity; a small amount could cover a large surface.

And about the alternatives:

Titanium white has far greater opacity and tinting strength than lead white, and it can easily overpower most other pigments if not mixed carefully. Titanium white has been criticised for leading to “chalkiness” in mixtures.

Zinc white dries slowly and creates a relatively inflexible paint film. Critics of the pigment argue that its use leads to excessive cracking and delimitation, even when used sparingly.

Why the hate for the great technology that is lead paint!?

Well, the answer is pretty simple: “Usefulness” and “harmfulness” are completely independent spectres.

I see this conflated a lot in discussions surrounding AI: People who are wary of the harm transformer models can cause, often also call them “useless hallucination machines”. And others, will respond to the claims of harm by pointing to the usefulness.

It’s obvious that the people and companies who’s developed these tools, have spent very little time thinking of the harmful consequences — and have only focused on making them useful. ¹

Perhaps not quite as obvious, is the way some critics hurt the discussion.

I think generative AI can be very useful — even in its current state. For instance, I can’t draw, but frequently think in images — so using image generators would be great for my blog. ²

I would rather not use the tools for writing (or translating) directly — but I do have some commands set up in Raycast to quickly check the spelling of a word, or some general grammar. And as a programming novice, I’ve gotten a lot of help from Raycast’s integrated chat — even though it sometimes does stuff like this:

Me: Do I have to do this to get the negative value?
calc(-1 * var(--my-variable))

Opus: Yes, if you need to use a negative value of a CSS variable, you can use the calc() function to perform the calculation. The syntax you provided is correct and can be used to calculate a negative value based on a CSS variable. (…)

Summary:
Using calc(-var(--my-variable)) allows you to dynamically compute negative values from CSS variables, offering flexibility in responsive or dynamic design scenarios.

Me: Wait, so I don’t need the -1 part? Do I need the calc at all? Can I just do -var(--my-variable to get the negative value of the variable?

Opus: You can indeed simply put a minus sign in front of the CSS variable to negate its value directly. You do not need to use calc() or multiply by -1 for this purpose. Here’s how you can do it: …

Well, that last part is straight up wrong. 😅 And so was the example of calc(-var(--my-variable)). But I’ve also managed to do and learn a lot more than what I would’ve been able to do without this tool. ³

Now, let’s imagine someone named Franz, who finds these tools demonstrably useful — but isn’t mindful of the harmful sides. Then he hears someone say: “Transformer models are problematic in these (*insert thoughtful criticism here*) important ways — and also, they’re completely useless.”

Franz knows the last part isn’t true — so why give any credibility to the first part?

I’m issuing two challenges:

Thanks to regulation (🫶🏻) we’ve mostly managed to ban lead paint. And I’m not bringing it up because I’m arguing that we have to do the same to generative AI. ⁴ I’m bringing it up because I have one challenge for the critics, and one for the advocates:

• For the advocates: Not all advancements are worth it. ⁵ And technology isn’t inevitable: Something can both be very useful and so harmful that we shouldn’t do it. So don’t shrug off ethical questions by pointing to general notions of “progress” and potential use-cases. Instead, maybe work on reducing the harm enough so that it is worth it? And then I’m talking about “society at large”, not just your investors. ⁶
• For the critics: If someone, arguing for the regulation of lead paint, also refused to acknowledge its usefulness, it would hurt the cause. So while it is very valuable to point towards areas where it’s a terrible idea to use AI tools, don’t let the advocates conflate the “harmfulness” and “usefulness” debates.