First question ...
...why the fuck do you need a Bing App for your fscking wallpaper ?
Can we please shift to another timeline, this one is clearly broken ....
216 publicly visible posts • joined 22 Oct 2007
I realize this isn't going to be a very popular post; downvote if you want but please hear me out. I'm in neither the "burn it now" nor "machine salvation" camp, nor do I want to 'convert' people ...I just would like people to understand and have a discussion instead of a sermon.
Let'em burn. To the ground. All of them.
Maybe we can sift through their ashes and find some specific applications where it will be useful. I don't think it'll go away entirely, I can certainly see places wher it would be useful ... on a much smaller, more limited scale - once wiser and cooler heads get together and find a way forward that doesn't run roughshod over society. We really need to find a wayu out of the data issue before either governments or other corporations get involved, because that's not going to end well for ANYONE.
While I think the tech overall is interesting, the hype slicksters, marketing bloviators and insufficiently sceptical commentators have blown it up way past anything feasible with current knowledge and computing ability.
At the core it's not "intelligence" but (pretty brute force) pattern matching ... and despite looking/feeling intelligent it's _not_, and it's not that terribly hard to hit edge cases or run headfirst into training data limits. And unlike humans, these systems cannot assume, invent or pull disparate associations together very well. (There is research underway to try to solve this, but it's still pretty early)
The term for it is "solution space" (How much room, and whether or not a model can 'solve' a given input for the data is has available) and it can be surprisingly small. Of course you can compensate for that by throwing more data and compute at it - which is exactly the solution now, but that stops being feasible fairly quickly.
The best example I can give of this:
Invent a mythical creature. If you go to an artist,-even a novice one- you can describe it and from that description, the artist can infer what it is you want and render it into a drawing.
But a model can't do that. You can prompt it, you can decompose it into pieces and features, but if it has no associated "knowledge cluster" that covers those concepts, you'll get nothing - or at best some kind of chimeric nonsense instead. More powerful models might do a slightly better job of it, but still run into the same problem.
Also casual fact: democratization is already well underway. A lot of models can be run on consumer grade hardware already, and perform pretty well within that environment.
LLM's might find a place as specific knowledge field query systems .. once you accept its limitations and you learn to shape questions, it's a good way to learn or synthesize/gain knowledge.
Text to image/image to image has a lot of potential as well -- for novice artists as a structural aid (via ControlNet), for experienced people as infill or emergent decoration.
Disclaimer: Yes I've worked with both .. and come to realize very quickly both are powerful tools - and all powerful tools carry an abuse risk, especially when they have such a low barrier to entry - but what's struck me the most is how little literacy there is surrounding all of it. A lot of what I read is almost caricaturial; and a lot of possibility assumptions are either on or beyond the edge of capability.
I apologize for a bit of a rambly post, I originally was just going to say one thing then realized I had to James Burke Connections style provide better context for people.;-)
Take care all,
... would be the grassroots AI community managing to jailbreak the model and enabling it to run entirely divorced and isolated from Google itself. Even better if they figured out how to (re)train it or interface it with other (existing) data. At that point, there's a wonderful poetic justice in Google trying to make people more dependent ... and giving people more independence instead.
You can draw the parallels to manufacturing:
Early manufacturing had huge machines with low efficiency and limited usability.
Over time, machines got better, more versatile .. and then then CNC revolution happened.
Right now you can already get CNC machines at the home level now (micro mill/lathe), to say nothing of how 3d printing has revolutionized DIY.
It's likely that eventually this will happen to LLM's as well.
Will it be a better world ? I have no clue. I can't (and won't) judge -- but it'll be interesting to watch.
I reluctantly migrated this week. Not because I wanted to, but to at least do it on my time and terms rather than to find out I have to rush through anything.
To keep W11 livable I used:
- Win11Debloat (get rid of all the crapware)
https://github.com/Raphire/Win11Debloat
- ExplorerPatcher (fixes the taskbar so you can actually move it to where you want, rather than where M$ tell you to)
https://github.com/valinet/ExplorerPatcher/
Do I want to move to Linux ? Yes, but I just lack the smarts and the will at the moment. Forgive me, I am weak.
... but I still am using WinAMP as my media player of choice. Classic skins somehow broke earlier this year from one day to the next (playlist window stopped opening), but the modern skin packages still function fine.
I've tried some of the modern fork/offshoots but none of them managed to capture the ease of use or the convenience of the original - so here's a toast to WinAMP.
Cheers mate, carry on and stay llama ass whoopin' awesome
Just as America and their allies (and a whole lot of special interest groups) are leveraging the same [social] media to inflame hatred of those "enemies" (or create entirely new enemies), relying on the average social media user being too stupid/lazy to think for themselves, blindly believe what they're being told and launch 'mostly peaceful protests/occupations' "for [insert cause here] justice" among other things ?
Now that I have your attention with that little snark (for which I'm only a little bit sorry) I feel the the need to point this out:
It's a VERY slippery slope between disinformation and dissent, and the dumber you keep/make people, the further you can shift that line without anyone realizing.
Do you want to fight disinfo/misinfo ? Give people perspective and encourage/teach them to critically think FOR THEMSELVES.
Don't try to 'control the narrative' and bombard people with the one side you prefer and exhort people to 'trust the experts' while discrediting/removing the other side.
And that requires access to information on ALL sides of ANY argument; political, economical, scientific, religious, societal, ideological - you name it.
I don't think anyone realize this is still censorship, albeit in a very alluring guise for a noble (corrupted) cause, and effectively, doing this means "the West" has ALREADY lost, emplyojng the same tactics we criticize the nations we call 'enemies' for.
I was going to post this anonymously, knowing it's a pretty spicy/contrarian post most will probably downvote to hell.
But please, before you do: think -for yourself- on what I've said and what it implies.
I have, and that's why I'm posting this under my own name.
Put nothing in "the cloud"
- that you don't have a certified, current and regularly checked, usable and updated (local) backup of.
- that is too important to lose (for whatever reason)
- that you must have access to at critical moments.
Completely
Lost
Our
Unmissable
Data
;-)
Because MS insists on shoveling features they haven't fully tested (if at all) at their hapless victims (otherwise known as "users") in a desperate bid to prove their OS is ready for primetime.
And rather than STOP and fix the mess they made last update (or any of the previous ones), they happily/blissfully move on to the new flavor of the day insisting that this time, they'll get it right!
They say insanity is doing the same thing over and over and expecting different results ... making the institutional mental health of the Microsoft edifice questionable at best, mythical at worst. Take your pick.