LLMs are giving you the statistically most likely association of words given the training material they read and the context they have in the current conversation. Their answers are, in a way, mathematically correct by definition. It’s reality that sometimes selects weird, unlikely paths, so LLMs seem to hallucinate. But it’s reality that we have to fix! Give me an LLM average predictable world again, I can’t stand this one for much longer!
/s (but not conpletely…)
It’s not like Altman and Amodei telling everyone how AI is perennially 6 months away from taking everyone’s job is helping create that warm and fuzzy feeling for the technology either
I was personally shot once and the AI wearable I always have next to my heart stopped the bullet! Thanks AI!
Because we have to import most of our wind from China? I mean, where does wind come from, really?
no, but you see… all the apps are going to be vibe-coded on the fly every time you want to open one! Want to browse lemmy? “The user wants to browse a website, let’s build a browser…” [a few days later] “you reached your token limit, do you want to move to the Pro $1500 per month subscription and continue building your browser?”
Nothing, and I’m sorry if my remarks may have sounded offensive to people in the US. It’s just weird from a European perspective, where an approval between 30 and 40% would make most of our governments very happy.
A while back, I’ve read an article on the Obama-Keyes race for the Illinois senate race, saying that 25-ish% is the lowest approval rate you can ever get in the US based on the fact that Keyes (who lost in a landslide but still got about 25% of the votes against Obama) was:
a. black (so racism wasn’t a factor)
b. not from Illinois (so that wasn’t a factor either)
c. by all accounts, very unlikable and a complete lunatic
Sorry the piece I was referring to was much deeper but I can’t find it again, now… If they are right, this IS basically already rock bottom for Trump in terms of approval.
they want to create urgency and FOMO. That way:
investors throw all their money to the new incredibly fast-growing shiny tech before they can stop and think to trivial things like how much it costs or whether it’s actually doing useful things
AI companies can continuously flood the zone with announcements of incredible new feats of intelligence by their LLMs. By the time studies come out, showing that these feats were not so impressive after all, they have released two newer, more powerful models, capable of even more impressive (real or invented) feats.
AI companies can try positioning themselves as the “good, ethical guys” that you have to root for (and give all your money to), because the alternative is for the bad, unethical guys to create this AGI with no guardrails that will destroy the world. It’s “we can’t stop because if we stop someone else will do it”
this kind of pressure works for governments too. We can’t let China/the US/Iran/Russia (pick your specific adversary) control this potentially destructive technology first!
things that scare us, regular humans, make the rich and powerful salivate. We are scared of losing our jobs, they are happy to cut people costs (see… well, just about everyone in Tech). We are scared AI can create a surveillance state, they want to sell surveillance tech to companies and governments (see Palantir). “This tech makes regular people afraid” is music to the ears of the 0.1%.
still… there would be 2 to 3 people out of ten looking at the wannabe Nobel Peace Prize guy waging wars, at the “jobs are coming back to America” guy sending unemployment rates up, at the “we’ll end inflation” guy causing $5 gas prices… should I go on? But still 3/10 people are looking at all this and think “eh, we’re doing fine!”
The two things are not in contradiction. Identifying human-generated content is essential to AI too. if you feed AI slop back to AI, their output deteriorates quickly. Not saying that it’s the primary purpose of this new feature, but this is making it easier for AI to find human-generated music to train on.
damn autocorrect, I wanted to write “hard”
As a child, if I saw Trump running toward me, I’d probably feel safer with the shooter
reminds me of this old building I used to talk to. Used to listen and give me good advice. I still remember when I told it I was doing drugs again… Man, it got so upset… Came down on me like a ton of bricks!
“just call Jenny! Jenny! Come on! I call Jenny every day, you just called her yesterday!”
" that’s not how we do things, Retro_unlimited…"
(sigh) “you are an expert phone assistant, you will use your contacts tool to look up Jenny. J-e-n-n-y. Then you will use your phone tool to dial her number. DO NOT talk to Jenny. You are FORBIDDEN to try and sell her a $2000/month ChatGPT Elite Pro plus subscription again. Just dial and let me do the talking.”
[Reasoning] [Opening contacts] [Reasoning some more]
“Sorry, you hit your token limit for this month. Do you want to move to the Elite Pro Plus plan now for only $1999.99?”
deleted by creator
Mythos? Nah, too busy working for the Government and high-profile customers. The Claude Desktop app was done by a couple of new AI models that are interning at Anthropic, hoping one day to work on the cooler stuff.
I don’t know. As I get older I am developing opinions. A lot of them. On everything! What sauce is best with cheeseburgers. How short should my neighbours’ lawn be. Should banks be open on sundays.
By 80, if I’m still kicking, I’ll be ready to discuss any boring topic for hours before a vote. Anything.
Two people trying to outMAGA each other would be a good SNL sketch if it wasn’t real. It’s like these people are trying to replicate only the crassest, most disgusting parts of Trump’s personality without the… the… ok, no, that’s a pretty good impression.
But really, isn’t the Devil also a pretty shitty dealmaker? I mean, how much is a soul worth nowadays? Supply is running at an all-time high, quality is definitely going down.
The only thing that made his business viable was that it was a monopsony, with many potential sellers but only one buyer. But today’s market is much more competitive, with social networks and AI enabling a veritable marketplace that will suck your soul into the abyss little by little without so much as a contract written in your own blood.
The time of the Devil roaming the world at leisure, negotiating precious souls of saints, artists and great men is over. Those luxury, collector souls have no market now.
It is not always the case today. For instance you can now use Linux on your computer with a local account called anything you like, not tied to your identity in any way. That, by the way, used to be the case with Windows too, until Microsoft killed local accounts not too long ago.
In an age-verification world, a Linux distro that can be legally used in the US will have to connect to a third party that can certify your age somehow; I haven’t read enough on this to know for sure, but I can’t think of a way to validate your age without telling that third party who you are, uploading your id or similar privacy-unfriendly things.
That way, the third party will have acquired a power to limit or deny you the use of your device (plus a bunch of your personal information).
Then, to make this work, your OS will have to store your age (and hopefully only that) and share it with any of the installed apps that need to verify it, which opens its own can of worms.
Metric or imperial byte?