Earlier I asked GPT how to export DXFs in Blender and it bullshitted me, so I asked Google the same question, and it gave me this site:


Which is entirely generated by GPT.

Similar bullshit, but this time it lucked out and got the right answer. At the bottom of the page there is a link to an almost identical site, but this one is about child-care (what could possibly go wrong?). It is now easier to write the code that can generate an infinity of websites than it is to write an actual website.

And they’ll all be bullshit but it doesn’t matter because there’s more money in gaming the algorithms than there is in actually creating anything of value.

Which is a direct and inevitable consequence of paying for communication with advertising.

There are also (increasingly sophisticated, and increasingly simple to use) services springing up that (for a small monthly fee) will generate bullshit for you – because what the world really needs is super-intelligent machines generating an infinite amount of advertising on the fly. Instead of turning the entire universe into paperclips it’s all about stealing attention. A machine for manufacturing desire. In the future, plants will have advertising on their leaves.

So Google has 2 problems:

1) People are going to be using GPT instead of Google to find answers because rightly or wrongly, it presents a needle rather than a haystack, and a lot of people are looking for confirmation rather than information.

2) Google is increasingly going to be indexing sites generated by AIs who are (at the time of writing) truth-agnostic but expert at gaming Google’s aglos.

We (the people) have 3 problems:

1) We are rapidly heading towards creating intelligences far greater than ours, who’s goals may not align with ours, and which could well drive us to extinction

2) These intelligences are being created and controlled by other non-human intelligences… ie: corporations, who’s goals definitely do not align with ours, and who are already driving us to extinction, and who rather than doing anything have decided to lie about it.

3) These intelligences are already in arms-races with each other, putting short-term gain before long-term survival.

So what do we do?

No idea, but the bottom-line probably has something to do with person-to-person networks of trust where not every single human communication happens on someone else’s computer.

A level up: Radical transparency for machines; Radical privacy for humans.

Anything that isn’t open-source is fundamentally untrustworthy.

Power asymmetry is the root of all evil.

Leave a Reply

Your email address will not be published. Required fields are marked *