A lot of people on HN are anti-overhyping, which comes across as being opposed to the thing being overhyped. It was similar when cryptocurrency overhyping was popular.
That's what I feel. Whenever I see someone writing something to the effect of “LLMs will replace X” (where X could be literally anything including software developers) I get an intense urge to write something against it. Not something nice, mind you.
I also find takes from the anti-LLM to be exceedingly dumb at times. Oh this text has this and that, it must have been written by LLM and thus is not worth even considering.
I'm personally amazed by what "A.I." is actually capable of, but I have a fairly solid understanding of what's going on "under the hood" of it, and therefore have somewhat realistic expectations of it. Then I see folks go overhyping it's capabilities because they've drunk themselves stupid on the lies they've been told about what it is and what it's capable of (and it's simply not capable of what the liars at the top of the A.I. corporations are telling everyone). Just try to temper their enthusiasm with a bit of reality and you're instantly "anti-AI" or "doomer" or some other just completely wrong characterization. At this point I'm convinced that for a lotta folks, A.I. is just another literal cult just like politics these days, or crypto-coins not that long ago... Drink that kool-aid, I guess... ~shrug~
I also don't think that many of the so-called "anti-AI" folks are so much against AI itself, as they're against the unethical ways that certain folks "at the top" are using it to do massive harm in an attempt to try to satisfy their bottomless greed and lust for power, and against the ways that some other folks are using it to basically escape the need to think at all, even when their job requires actual thinking.
Sounds like me. It's awesome. Only the sheer amount of very clear fraud and the way an entire way of people who, not too long ago, where barely functional, have jumped on the bandwagon (same as with crypto) just sours me on it.
That and Jenson screwing those of us who made him over the last decade.
It'll be better once the fraudsters are in jail and once we're able have boxes with ~1TB ram running off of our solar in the garage.
Well... It's got that potential, that's for sure. It could pretty easily be pushed straight past that line into "awesome" if not for those absolute clowns "at the top" holding it all back decades behind where it should be, all to play their little game of "He who dies with the most money wins" at the eventual expense of all life on Earth (at the rate they're going). It's just so sad the petty misuse and waste of resources the ultra-richest of the rich are choosing to be. Actively working against the betterment of themselves and the world around them, for the most insane reasons, when they could be using all this amazing tech to build a genuinely better world for themselves and everyone else.
Honestly, I look at where technology's gone since my early days (300baud acoustic phone modems, 64kilobytes of RAM, 1Mhz 8bit CPU days) being utterly fascinated by how actual science was so quickly catching up with science fiction (Star Trek being one of my bigger influences in that area of interest), and I see so many truly amazing things that we've invented / built along the way; if we were a more cooperative society instead of "Law of the Jungle" hostile greed-driven society, we'd already pretty much have that Star Trek reality today (minus the faster than light travel bit, as that's apparently "impossible" according to current theory and math AFAIK).
We've already got a bunch of Star Trek level tech, and we could have most of it I suspect. 3D printing? Not terribly far off from ST: TNG "replicators". LLMs? Not too far off from the Star Trek computer interface. Smart Phone? Pretty much a Star Trek PADD (tablet computer). Tricorder? Well... Smart Phones are gettin' pretty close. A few more fancy sensors would pretty much do it. Holodeck? Well, that one's a bit more tricky, but who knows where VR would be today in a society that wasn't totally 100% beholden to the cult of money?
Oh absolutely! This made me wonder and there was an exact post with similar title but instead of AI it said Crypto
Ask HN: Why is Hacker News so anti-crypto? : https://news.ycombinator.com/item?id=31302494 (Do note that the post is flagged and there might be some good moderation reasons for that)
This is the reason that most of us at HN might dislike overhype. I have seen a lot of these crypto users move from crypto hype to AI hype.
Every few years, people forget the last shiny thing and move to the next and think why is X crowd not invested in Y? They must be anti-Y!
The hype was totally justified for AI now that coding has been completely transformed for the most part.
Trying to “correct the hype” just looks like clowning in hindsight.
It’s like people in 1990s were trying to correct the internets hype: oh you know it won’t change anything and the tech bros want to create hype out of nothing!
So it’s not neutral to be anti hype for AI. It is just wrong.
Those of us who were actually there in the late '90s also remember the insane overhype, grifting, and fraud that led to the dotcom bubble and then the painful dotcom bust afterwards that took the entire industry years to recover from. That pattern is playing itself out again.
Those who analogize today's AI frenzy to the internet adoption frenzy of the '90s in a positive light abjectly fail to understand what an incredibly bad look that is.
I'm in my mid 40s and I'm now switching between three different pairs of glasses for different distances. Bifocals and progressive lenses wouldn't help given the things I need to focus on are usually directly in front of me. Changes have been happening rapidly.
Interesting. Research results seem mixed, and I wonder if ASD comormidity is affecting them. https://pmc.ncbi.nlm.nih.gov/articles/PMC7369237/ has a reasonably large sample size and finds an effect for ASD but not ADHD.
May be a timing coincidence - I'm in my mid 40s and my eyes are changing rapidly and getting much less flexible in terms of range. I'm switching between three different pairs of glasses for different distances, and need new prescriptions every year.
Persian Gulf has 20% more salt in water because of the humans which are throwing the oversalinated waste back into the sea. Dehidrated salt may be a big deal for some areas because of no waste into input.
Through the magic of Googling "Persian Gulf salinity" it seems like it's more that it's a shallow Gulf in a dry area so it has significant evaporation. Desalination does effect it but it's only a few percent of the total evaporation (which is still surprisingly big) and doesn't sound like the main driving factor or an imminent ecological concern.
Look at a map. The Persian Gulf is a dead end, and all ocean water flow has to come through the Strait of Hormuz. There's some fresh water coming in from the Tigris and Euphrates rivers, but less each year as that fresh water is captured and used, and as global warming increases evaporation.
The San Francisco bay has to be actively managed for similar reasons. It's a large body of water with a narrow outlet, fed by a river system from which much water is captured. If too little water comes in from the Sacramento River, the delta will turn to salt water. Managing that is what the Bay Model, mentioned recently, is for.
If salt and water flow in but only water flows out you will be left with salt. Same reason that concentrated brine comes out of a desalination plant, or that the dead sea is what it is.
normally your browser should translate the page if not let me check it and overwork it within a few days.
goal of the simulation are 2 things
at first i want to expand the system to lay a foundation for simulating emotional behavior by AI. Actuall LLM can act emotionally by prompts but dont get what emotions do mean nor they do have an inner state. they just roll chances in hope of fullfill theire task.
second goal I have to bring the simulation tto a point where you will have NPC2.0 for example non scripted NPC merchant who dont open his shop cause he is sick or he just really renember that the player stolen from him. It sad reality that NPC behavior havent really evolved for at least 20 Years
ps
to the free energy principle the theories from prof Friston explaining if I explain it very easy, that the human brain do not like surprises and for minimize the surprise it try to predict. This principle you can use for every kind of emergent behavior. best example from last houres. Some user tried to teach the agents the concept of PI wich is pretty hard cause I never coded them math knowledge but over night at least one agent figured out that PI is somehow conneced to 3.141 fuher more this agent weighted this knowledge on his own will as importent so he stored it in his longterm memory
The white paper says "mean-active memory pressure down to 1.95 GB for 1-bit Bonsai Image 4B and 2.38 GB for
Ternary Bonsai Image 4B". Storage is on the linked page, and is about half that.
That is very low, looks like it should run in base MacMini M4 with 16GB RAM. I understand it is not released yet? What sort of harness is necessary for this type of model? (I have only used coding agents through GH Copilot in VS Code, the JetBrains AI tool and Pi, this last one was sort of a pain to setup…)
reply