Hacker News Reader

> However with LLMs, there is one major worry - in that it encourages de-skilling and getting addicted to having a LLM think for us. Now, if you can run your own LLMs, youre resilient to that.

Not sure what you mean. Using a local LLM or a cloud one can get you addicted in the same way?

View on HN

Using a local LLM opens all the guarded secrets bare.

For example, you can run multiple LLMs to understand each outputs.

You can issue system messages (commands) to do specific actions, like ignore arbitrary moralities, or to process first and second derivative results.

By running and commanding an LLM locally, you become a actual tool-user, rather than a service user at the whim of whatever the company (ChatGPT, etc) wishes.

Where do you do recommend someone start the journey of being a llm service user to a llm tool user?

What kind of hardware is needed to get something useful running locally?

Reply on news.ycombinator.com