AmbitiousProcess (they/them)

  • 0 Posts
  • 4 Comments
Joined 3 months ago
cake
Cake day: June 6th, 2025

help-circle

  • That’s not an extension cable, but an adapter, thus it’s not a problem in this case. It’s a cable that can convert the data from an audio jack to something that can go through USB-C, not a cable that simply extends a USB-C cable. The cable can almost certainly handle any amount of power and data that an audio jack would pass through it, no problem, even if it were a USB-C to USB-C extension cable, and not an adapter.

    The problem arises when someone tries using a higher-spec USB-C cable with a lower-spec USB-C extension cable, such as using a 240W charger with the lower-spec USB-C extension cable in the middle that can only do 120W. In that case, it would pass more electricity through than the lower-spec cable could handle, and it would overheat.

    The amount of data and power from an audio jack is simply too small to overwhelm practically any USB-C cable or adapter that exists, thus it’s not an issue.


  • Most of these AI crawlers are from major corporations operating out of datacenters with known IP ranges, which is why they do IP range blocks. That’s why in Codeberg’s response, they mention that after they fixed the configuration issue that only blocked those IP ranges on non-Anubis routes, the crawling stopped.

    For example, OpenAI publishes a list of IP ranges that their crawlers can come from, and also displays user agents for each bot.

    Perplexity also publishes IP ranges, but Cloudflare later found them bypassing no-crawl directives with undeclared crawlers. They did use different IPs, but not from “shady apps.” Instead, they would simply rotate ASNs, and request a new IP.

    The reason they do this is because it is still legal for them to do so. Rotating ASNs and IPs within that ASN is not a crime. However, maliciously utilizing apps installed on people’s devices to route network traffic they’re unaware of is. It also carries much higher latency, and could even allow for man-in-the-middle attacks, which they clearly don’t want.


  • While true to a degree, I think the fact is that AI is just much more complex than a knife, and clearly has perverse incentives, which cause people to use it “wrong” more often than not.

    Sure, you can use a knife to cook just as you can use a knife to kill, but just as society encourages cooking and legally & morally discourages murder, then in the inverse, society encourages any shortcut that can get you to an end goal for the sake of profit, while not caring about personal growth, or the overall state of the world if everyone takes that same shortcut, and the AI technology is designed with the intent to be a shortcut rather than just a tool.

    The reason people use AI in so many damaging ways is not just because it is possible for the tool to be used that way, and some people don’t care about others, it’s that the tool is made with the intention of offloading your cognitive burden, doing things for you, and creating what can be used as a final product.

    It’s like if generative AI models for image generation could only fill in colors on line art, nothing more. The scope of the harm they could cause is very limited, because you’d always require line art of the final product, which would require human labor, and thus prevent a lot of slop content from people not even willing to do that, and it would be tailored as an assistance tool for artists, rather than an entire creation tool for anyone.

    Contrast that with GenAI models that can generate entire images, or even videos, and they come with the explicit premise and design of creating the final content, with all line art, colors, shading, etc, with just a prompt. This directly encourages slop content, because to have it only do something like coloring in lines will require a much more complex setup to prevent it from simply creating the end product all at once on its own.

    We can even see how the cultural shifts around AI happened in line with how UX changed for AI tools. The original design for OpenAI’s models was on “OpenAI Playground,” where you’d have this large box with a bunch of sliders you could tweak, and the model would just continue the previous sentence you typed if you didn’t word it like a conversation. It was designed to look like a tool, a research demo, and a mindless machine.

    Then, they released ChatGPT, and made it look more like a chat, and almost immediately, people began to humanize it, treating it as its own entity, a sort of semi-conscious figure, because it was “chatting” with them in an interface similar to how they might text with a friend.

    And now, ChatGPT’s homepage is presented as just a simple search box, and lo and behold, suddenly the marketing has shifted to using ChatGPT not as a companion, but as a research tool (e.g. “deep research”) and people have begun treating it more like a source of truth rather than just a thing talking to them.

    And even in models where there is extreme complexity to how you could manipulate them, and the many use cases they could be used for, interfaces are made as sleek and minimalistic as possible, to hide away any ability you might have to influence the result with real, human creativity.

    The tools might not be “evil” on their own, but when interfaces are designed the way they are, marketing speak is used how it is, and the profit motive incentivizes using them in the laziest way possible, bad outcomes are not just a side effect, they are a result by design.