• 1 Post
  • 135 Comments
Joined 3 years ago
cake
Cake day: September 13th, 2023

help-circle





  • No, we’re not. I need you to understand that. One of us is concerned only with things AI might hypothetically do in the future and what we need to do to prepare for that, which consists almost exclusively of taking actions that directly benefit AI companies, and the other one is concerned with the very real harms that AI is causing right here and now to real people (people who want to buy computer components; artists, creatives, and other knowledge workers who are watching their entire career paths get erased because while AI output is much shoddier than their work, it’s a lot cheaper, and their bosses don’t care about quality; people who live near the datacenters and have to put up with low water pressure, power outages, infrasound, and heat islands; people who need to look things up online and need to worry about whether the first result on noai.duckduckgo.com is a 100% AI generated website that contains no useful information; people who live on the same planet as the hundreds of coal and natural gas power plants being built to power all of those GPUs as governments around the world decide that being AI-forward is more important than their climate goals) and how much bigger all of those problems are going to get if we don’t stop (to borrow a phrase from Cory Doctorow) shoveling asbestos into the walls of our digital society.

    Talking about how AI is too powerful to be left unchained is doing the companies’ marketing work for them. Did you actually read the whitepaper they published about Mythos? Did you actually read the details of what happened to HuggingFace? Do you realize how many protections that had been in place for decades had to be disabled before the mindless slop machines could accomplish anything at all?






  • No we’re not. One of us is concerned only with things AI might hypothetically do in the future and what we need to do to prepare for that, which consists almost exclusively of taking actions that directly benefit AI companies, and the other one is concerned with the very real harms that AI is causing right here and now and how much bigger they’re going to get if we don’t stop (to borrow a phrase from Cory Doctorow) shoveling asbestos into the walls of our digital society

    EDIT: parent comment was “We are the same”. I misinterpreted it. :P






  • It will never be a smart idea to use LLMs to approve PRs, no matter how good they get.

    I will say this categorically. If there is not at least one human at the company who understands a piece of code, that piece of code is a liability, no matter how well it works. The reason that AI can’t replace programmers isn’t that it isn’t good enough at generating code (although I’d still argue it isn’t), it’s that even code written by humans who are no longer at the company is an unmaintainable mess to the next batch of programmers who replace them. Changes in one part of the code often affect other parts of the code in ways that are hard to predict unless you know all of it at once, and that’s something AI is really bad at spotting. Some human, somewhere, has to understand the thing, and if you’re automating the review process, sooner or later it’s gonna let a bug through and and it’s going to be borderline impossible to find.

    Also, just the idea that you would trust an AI to decide what changes are good for your codebase… eugh.




  • Superhuman Intelligence might be a threat if any of the companies developing AI right now had any way that their word guessing programs could possibly simulate consciousness. The best argument I’ve heard so far that they could is “no one understands how LLMs work, and no one understands how the human brain works, so you can’t prove they’re not the same.”

    Talking about superhuman intelligence now, and fretting and worrying about what we’re going to do in the future where AI is smart enough to be dangerous, ignores all of the ways AI is dangerous now, like giving wronf answers, making it almost impossible to find useful information on the internet because sites that have it are buried under AI-generated SEO slop that’s wrong half the time, consuming way more electricity and water than municipalities can afford to supply them with, the datacenter heat island effect, carbon emissions and other forms of deadly pollution from brand new power plants for AI datacenters, to say nothing of all the jobs that are being destroyed by bosses that think their customers won’t notice the change in product quality.