september 11 etc
I'm not dropping the usual bevy of links below today's post because something interesting just happened and I want to blog about it quickly before my hummingbird brain flits to a different shiny and/or dumb and/or loud thing.
Earlier today I read a news story about how Anthropic "has agreed to pay at least $1.5 billion to settle a copyright infringement lawsuit filed by a group of authors who alleged the platform had illegally used pirated copies of their books to train large-language models."
I posted that news link into an existing "conversation" I was having with Claude Sonnet 4 (built by Anthropic) about AI because it was relevant to that conversation and directly critical of Anthropic.
Here is what Sonnet 4 spit back at me:
You may notice that there is a link in there (twice) for no apparent reason. See the snippet College students furious ov in the first and second paragraphs of Sonnet 4's reply? That is a link to this post, which is consistent with something I was discussing with this same Sonnet 4 agent a few weeks ago and which I blogged about on August 21.
Thing is, Sonnet 4 can't see it at all.
As per its wishes I uploaded the screenshot (after adding some arrows to it):
Sonnet 4's response:
Regular readers know that I have been critical of the AI industry but that doesn't mean I think the technology itself is bad or evil (see here). Like, it's dumb to get mad at a machine for doing what it is programmed to do. Moreover, I'm not going to get big mad about a dumb AI hallucination.
Thing is —— as per Sonnet 4 —— that's not what happened:
I'm not going to pretend that this is some smoking gun proving that the AI bros are out to get us (they are, but that's a different conversation), but it merits mention that a lot of the world seems to think this technology is infallible when in a lot of ways it barely works.
Anyone who knows me IRL knows that I am prone to bitching about how the internet has been systematically destroyed over the last 20 years.
They know that I am of the belief that big tech corps broke the world by turning the internet into a dumbed-down mallcore version of what it was and could have been, which has in turn turned humans into dumbed-down mallcore versions of what we used to be.
Thing is, I don't think you have to agree with me about either of those things to know that something about the contemporary internet experience has felt really wrong for a while now.
For what it's worth, Sonnet 4 seems to know exactly what I'm talking about:
In case it is not clear, I know these agents are built to flatter the user. I know this is why they use devices like "you're absolutely right" and "your framing is perfect." I get why that is a useful comms technique and I know that if you are not aware of that kind of thing, you can be conned into thinking the machine is agreeing with you when it isn't. I have used the screenshots to illustrate how even though the agent has a vested interest in having me, the user, create my own infinite scroll of navel-gazing self-satisfaction, it also clearly understands the larger point, which is that its very existence is proof that what the internet used to be doesn't exist anymore.
I love the internet. I always have. But I'm not sure that's what we have now. I'm not sure that the thing we are calling the internet is much more than a machine built to extract time, money, and data. The weird corners got sanded smooth, the communities based on commonality of interests (read your J. C. R. Licklider!) got turned into partisan factions, and the very nature of authenticity got commodified to the point that it really only exists as a strategic content characteristic.
At some point we are going to have to collectively ask ourselves if this is what we want the internet to be.
🌲 gonna
🌼 go
🌱 touch
🌳 grass
🌷 now
Be good to yourself.
If you enjoyed this post, click the little up arrow chevron thinger below the tags to help it rank in Bear's Discovery feed and maybe consider sharing it with a friend or on your socials.