• 0 Posts
  • 23 Comments
Joined 2 years ago
cake
Cake day: July 5th, 2023

help-circle
rss


  • If you first have to write comprehensive unit/integration tests, then have a model spray code at them until it passes, that isn’t useful. If you spend that much time writing perfect tests, you’ve already written probably twice the code of just the solution and reasonable tests.

    Also you have an unmaintainable codebase that could be a hairball of different code snippets slapped together with dubious copyright.

    Until they hit real AGI this is just fancy auto complete. With the hype they may dissuade a whole generation of software engineers picking a career today. If they don’t actually make it to AGI it will take a long time to recover and humans who actually know how to fix AI slop will make bank.


  • The approach of LLMs without some sort of symbolic reasoning layer aren’t actually able to hold a model of what their context is and their relationships. They predict the next token, but fall apart when you change the numbers in a problem or add some negation to the prompt.

    Awesome for protein research, summarization, speech recognition, speech generation, deep fakes, spam creation, RAG document summary, brainstorming, content classification, etc. I don’t even think we’ve found all the patterns they’d be great at predicting.

    There are tons of great uses, but just throwing more data, memory, compute, and power at transformers is likely to hit a wall without new models. All the AGI hype is a bit overblown. That’s not from me that’s Noam Chomsky https://youtu.be/axuGfh4UR9Q?t=9271.




  • The fingerprinting I’m talking about gets encoded in the screen recording too. Subtle pixel changes here or there over the entire length of the video. It’ll be lossy when it’s transcoded, but over the whole video it’s there enough times it won’t matter. Even scaling to lower quality won’t fix it and then it’ll also be lower quality.

    It’ll be like DRM, there will be people trying to remove it like anything else. They’ll break one thing and another will come along. There would still be a black market, but most people can get an unrestricted copy in exchange for money so there’s one less reason to pirate.

    Unless you’re actually pointing a camera at the screen, then OK, you do you.











  • The headline stat is a misinterpretation of the study which was done by Arkose Labs which “provides businesses with lasting bot prevention and account security by sapping the financial motivations of cybercriminals.”

    That’s pretty vague but skimming it sounds like they prevent automated account creation and takeover. The stat comes from the companies they have access to (who need bot protection enough to pay for it), and 76% of activity on the login/account creation was malicious. That makes a lot more sense. All the various hacks and credential leaks result in bots banging in stolen credentials on high value sites.