Agreed, except I’m not frustrated by AI being overloaded-- it makes sense to me that a well-known, two-letter acronym be highly inclusive-- it’s frustrating that market dominance and marketing can just take over that acronym. LLM is a more accurate description than AI, but it would hurt their sales if most people understood it’s a statistical model and didn’t mistake it for a kind of intelligence.

I’m starting to think that those of us who understand this will probably fair better by forfeiting the semantics battle and making the distinction between LLM-AI and ML.

I think The Lorax comparison is warranted for LLMs, hyping the race for AGI, and AI as marketing/branding, but the AI umbrella includes a lot of Machine Learning that’s just practical, applied statistics.

A lot of ML applications aren’t anymore resource intensive then using Lemmy. So I think the discussion about harms v. benefits of ai-- beyond hypers and doomers-- often diverges from conflating “AI” with these AI companies that are all-in on LLMs vs AI as a field that contains an over-hyped, over-invested, resource-intensive, subcategory bubble that needs to pop already as well as subcategories that make it easy and efficient to classify and cluster data.

Slightly better is still the direction we want to head in. Not sure how else we get off the racketing-effect/boiled-frog path we’ve been on.

If you’re interested, the linux unplugged podcast had a recent episode on their experiences. I’ve never loved the tone of “hey, look, this is inevitable” when it comes to ai, but I can see its utility when well-scoped with conservative permissions and oversight vs letting it loose or vibe coding. Now if only hardware wasn’t artificially inflated I might think it was worth dabbling locally.

Sweet, I added autological word to my vocabulary just in time to use it.

I imagine this is how it’ll work for stage 2 of Ai enshittifation. They’ll just add a bunch of garbage upstream about a brand or product marketers are paying to push and it’ll infect a bunch of outputs downstream.

Lol, hittin the weights like Scratchy

Haven’t read the article and have a limited knowledge of ai, but I wonder if they do this for reinforcement learning: So OSS PR responses can be used to label different weights and models. Using even more free labor to train their models.

I wonder if hierarchical structures need reframing rather than removing. If changing our mental model could be the dismantling. I’m considering the definition and observation of emergent and beneficial hierarchies as discussed in “Thinking in Systems” by Donella Meadows-- the hierarchy structure is not inherently bad. What’s bad is, when it comes to human social structure, the person coordinating a collection of people is often considered more important.

If they were equally as replaceable as anyone in the collection (as it should be in a resilient system)-- perhaps by randomly reappointing that position, periodically-- then you could have a central-coordinator structure where benefitial, without the problems of that coordinator becoming drunk on power.

Coincidentally, that book has a quote from Zen and the Art of Motorcycle Maintenance that’s very fitting for that last part you mentioned:

if a factory is torn down but the rationality which produced it is left standing, then that rationality will simply produce another factory. If a revolution destroys a systematic government, but the systematic patterns of thought that produced that government are left intact, then those patterns will repeat themselves in the succeeding government. There’s so much talk about the system. And so little understanding.

https://ente.com/blog/self-hosting-quickstart/

Wouldn’t a byte be $2 if a bit was a quarter, or do you mean 2 bits are a quarter? Also i think you were right to use powers of 10 in your estimate. Article says kilobyte, not kibibyte. I really like what your conversion illustrates, I’m just tripping up on the details. I could be wrong-- commenting so someone can correct me if i am-- if a bit is a quarter, 69 Kilobytes would be $138,000

Does it have the model in the model?

Also reputation: former classmates or colleagues might be working there already

F - tier Only time I flew with them I had a layover. 1st flight was late. My wife was visibly, 7 months pregnant. Hurried to the connection’s gate. They closed it as we approached. People waiting in line to board on the other side of the gate, about 20ft away. Wouldn’t let us board.

No flight till 24hrs later. Said they’d cover the hotel but never got us vouchers. Had to get the credit card company to dispute the charge. Next day they charged us for carry on that was free on the first flight. Trip was for my grandma’s funeral. Made the funeral but lost an entire day with extended family.

IDK if I’ve ever hated a company or it’s staff more than that. So many layers of incompetence from different staff. Clearly a systemic problem.

During the pandemic I had the same back-to-back restriction for weed because I’d notice my mood dip after 2 days of use.

I appreciate your honesty, and I’m not far behind. I’ll scratch off build up sometimes and easily go months. Washing it is probably a similar frequency to getting my haircut.

Asking because the tumbler I use has a lot of little parts, so I easily go weeks between washing. I’m hoping someone more knowledgeable than me shares a good reason why I shouldn’t do that, because I feel a bit self conscious about it, but not enough to frequently wash.

Afaik you’d have to open a port and port forward for that to work, and you’d have to update every time your ip changes, unless you have a domain linked to it. There’s lots of other configurations, too: VPN/tailscale or equivalent onto your home network, a vps, reverse proxy, etc. I’ve yet to decide how to access from outside my home. Still tinkering locally, but mumble would be fun to try one day.

This is a handy meme to have these days. I hope to see it in circulation and to spread it myself! Thanks!

Careful on that high horse. When you become dependent on a tool you never owned it’s going to be a long way down when they inevitably enshittify. Willing to bet a lot of people in this community don’t even have an issue with some Ai assistance, but rather the hype, overuse, and laziness that separate “ai” as marketing from machine learning as a tool.