Spyke

Replies

Comment on

AI’s Memorization Crisis | Large language models don’t “learn”—they copy. And that could change everything for the tech industry.

Reply in thread

Nahh, even before LLMs became big, search results were becoming worse and worse because of all the SEO-spam.

Now you can generate coherent-sounding articles using LLMs, making the amount of trash even bigger.

Google making their search worse would be dumb, since all these LLMs also rely on it to some degree.

science

Comment on

Carbon dioxide overload, detected in human blood, suggests a potentially toxic atmosphere within 50 years - Air Quality, Atmosphere & Health

Reply in thread

I see this sentiment quite often, is it field-specific? Cause in physics and chemistry, starting at 0 is really not required...

In this case, the zero value is really not relevant (since no-one would ever have it anyway). It would just hide the signifcant drift over time. A good scaling here would be based on some clinically relevant interval I guess.

Comment on

A cool guide to move away from big tech and discover new and better tools!

Reply in thread

  • Step 1: have an old laptop/computer/Raspberry Pi that you can leave on at all times. Probably install Linux on it. Specs really do not matter, basically anything will work.
  • Step 2: think about what you would want to self-host. Check here for a good list. Start out with something that is not a real problem if it does not work, like a personal wiki or notes or something. Choose a reasonably popular project so you can look up problems more easily.
  • Step 3: try to install it: find step-by-step instructions and just follow them. Try to understand what you are doing. Ideally, use Docker containers for easy and reusable things.
  • Step 4: use it and see what happens!

Check the lemmy selfhosted community (or shudder its Reddit equivalent) to get some feeling for the community.

You reached the end