Absolutely needed: to get high efficiency for this beast … as it gets better, we’ll become too dependent.
“all of this growth is for a new technology that’s still finding its footing, and in many applications—education, medical advice, legal analysis—might be the wrong tool for the job,”
What benefit for society does this crop of large language models and generative “AI” offer?
We already see students use it for homework, meaning they don’t learn their stuff.
We also see people treat the output of LLMs as gospel truth, despite the fact that LLMs often hallucinate complete BS!
LLMs and generative “AI” rely on stolen artwork. Which is a net negative for society.
Some people think you can use it as a replacement for therapy or to fight loneliness. Turns out, simply reading fiction is better.
https://neurosciencenews.com/reading-emapthy-loneliness-28972/
Cool article. Thanks for the link. As an older fellow, but a lifetime reader, mainly of sci-fi, I feel thankful that I’ve been doing something to offset the years of drug and alcohol abuse. :) I’m considering joining a book club, now.
Recently, though, I went through and read all the novels of Robert Rankin up to ‘The Mechanical Messiah and Other Marvels of the Modern Age’, which kind of broke my mind a bit - I DO NOT recommend it, there’s like 50 of them - and it’s taken me a year or so to feel like starting to read novels again. They’re wildly entertaining if you’re a fan of running gags and sheer insane premises, but I shouldn’t have taken them all at once.
It’s pretty cool how books are more than just fuel for imagination, no? But I second the idea of joining a book club, because not only do you get the cognitive effects of a book, but you get the social benefits of a club!
Well for me, I enjoy pair programming my own projects with offline models. I also bounce ideas off it to attempt to ground myself in some type of reality (some models are better for this than others… probably has risk of delusions of grandeur. Some models will just verbally suck you off which is annoying).
I built ansible tooling for deploying k3s kubernetes and Ceph-backed Proxmox clusters and VMs and containers and services. Utilized the help of LLMs to structure my playbooks and figure out how roles work.
I love learning new things and LLMs have a lot to offer in that regard. You have to watch out for the bullshit and independently look at other sources as well, but it’s a great starting point and I can sometimes have sone deep conversations around some topics.