I also used it for advice on a massive personal decision, but I specifically asked it to debate with me and persuade me of the other side. I specifically prompted it for things I am not thinking about, or ways I could be wrong.
It was extremely good at the other side too. You just have to ask. I can imagine most people don't try this, but LLMs literally just do what you ask them to. And they're extremely good and weighing both sides if that's what you specifically want.
So who's fault is it if you only ask for one side, or if the LLM is too sycophantic? I'm not sure it's the LLMs fault actually.
> There’s no democracy in being mostly beholden to a few companies which own the largest and most powerful models, who can cut you off at any time, jack up the prices to inaccessibility, or unilaterally change the terms of the deal.
LOL. Maybe you are referring to OpenAI and Anthropic? Yes they have codex and opus. But about 1-2 months behind them is Grok, Gemini, and then 2-3 months behind them are all the other models available in cursor, from chinese open source models to composer etc.
How you can possibly use this "big company takes everything away" narrative is ridiculous, when you can probably use models for free that are abour 2 months behind the best models. This is probably the most uncentralised tech boom ever.
(I mean openAI is in such a bad state, I wouldn't be surprised if they lose almost their entire lead and user base within 6-12 months and are basically at the level of small chinese llm developers).
I agree with your general point, but there's an important qualifier:
> when you can probably use models for free that are about 2 months behind the best models.
You can use them for free, but training of near-SOTA foundation models is currently not an open-source process, and it's funded almost exclusively by large and/or wealthy corporations. That's a weak point at the moment from the perspective of openness.
> shed partisan tendencies and critically review policy initiatives
You could take a good will attitude to DOGE then. I think many (including Elon) genuinely believed they could cut fraud and waste. But by their own admission, they were only mostly an advisory committee.
You can only do so much. Congress still has authority, and that's how it works, that's how the system is intended. And the reason DOGE hasn't done much is exactly because congress isn't willing to cut spending. It NEVER will. It didn't under any president including Reagan.
So basically you have an ever increasing deficit and spending because the way the political system is setup drives this. In fact, it happens in basically every democracy, so maybe it's just something that happens in democracies.
So - you could call the promise of DOGE lies, but I think they were a lie from Trump and not Elon. I think Trump promised Elon cuts, to get his help in the election, then backtracked, and that's exactly why Elon stormed out, he didn't get what he wanted.
And the US government is still massively overspending. Trump didn't really cut anything.
Yeah, I remember people saying that about making 1m model 3s per year, landing rockets, getting 10k+ satellite privately into orbit, and getting millions of subscribers using internet via those satellites.
Maybe just maybe the guy does actually get things done, and if you didn't hate him you'd see that?
(yes, there are some things he hasn't gotten done. That doesn't take away from what he has gotten done)
Please understand that his companies succeeding in some things doesn’t make the things that are exaggerated, overpromised, or just plain naked hype with no backing somehow practical. It’s an interesting effect of our age that for some figures to some people if any criticism is considered unwarranted then all criticism must be disregarded.
It reminds me of growing up in the evangelical church and all the pastors who’d still keep their followers even after they show up in new cars or fly first class, taking the tithes from old ladies on their pension.
That's really understating things. He has promised so many things at various times that the "hits" are at best 10% of what he says. You can't just cherry pick his successes and say "well maybe this will work too" with a track record like that.
This mofo threw a Nazi salute and danced around on stage like an idiot with a chainsaw. Then he illegally downloaded the entire US treasury payment database and ran it through his AI and faced zero consequences. After promising to find a trillion in fraud and abuse, he left after less than half a year and declared there wasn't that much fraud after all.
To most normal people this long history of overblown claims and complete failures would disqualify him from serious consideration. To most normal people, a massive illegal siphoning of US government data would be beyond the pale and worthy of jail time.
But in today's age, there's enough smoke and mirrors that such a charlatan can just float on a sea of adulations right on past any consequences.
I actually think the public markets have a lot less faith in openai than softbank does. They need these crazy investors. Public markets would not value openai at $1.4trn. So they can't go public. It would reveal how bad things are.
He hasn't offered any production services to third parties whatsoever (seems to end at patents/standards gifting) - I wouldn't expect a fab open to third parties at all - if anything it'll be reserved for tesla/space-x/x.ai usage and candidly I don't think there's enough demand there to justify the capex involved in a fab.
I strongly suspect some sort of fab built by Elon associated companies will be announced soon. Almost all supply can be bought by Tesla and xAI.
Fab costs follow Rock's Law which is that the cost of building a fab doubles every 4 years.
Right now, it costs TSMC roughly $25-$40b for each N2 fab. It's going to cost $50b - $80b for an A14 fab 4 years from now. And then $100b - $160b again.
You need many customers in order to justify the cost. Tesla and xAI themselves won't be enough.
Standing up a modern-node EUVL fab is harder than rocket science. Tesla and xAI don't have the capex to do it, and the silicon they manufacture would be competing against much more experienced fabs like Samsung and Intel.
Frankly it is hopeless, I would be dumbfounded if Elon ever walked through the doors of ASML.
And so can you run local models which can generate far worse material.
And horny teens have always fantasist about celebrities, or that girl they have a crush one etc. And there always had been people people cutting physical images together to place the head of their obsession on some erotic magazine sourced body.
But like you saied it's creating a feed of all the people which have been sexualized against their will.
There is a huge difference between someone doing something in your mind (or room) and it staying there and it being posted international for billions of people to see (and download, and re-post, and cherry pick preferred pictures and then feed into AI model which will actually full undress people etc.)
and a huge company making money from not just sexualizing people against their will, but also putting creating a public feed about all the people they have sexualized against their will
and then the owner going out of their way to claim that that is all free speech they won't change anything and anyone who tries is fascist, communist, evil etc.
except that definitions of what free speech means vary largely between countries and huge parts of the world have definitions where stuff like "creating sexualized images of people against their will" (or systematic harassment, cyber mobbing, death threads, and a bunch of other things) are very clearly _not_ covered by free speech.
realistically speaking this is also AI output, i.e. not speech of a person (weather natural or legal/company), i.e. it's questionably if Grok posting generated images does even count as speech (in the US and many other countries)...
You know how the NHS reduced waiting lists a few years back? If you had waiting lists of say 100 for a surgery, they basically said - the list is maximum 15 people, after that it's whoever books first who gets the surgery. So basically you had to be lucky and be the number 15 on the list once a spot was open.
But! Magically NHS waiting lists got shorter! The government could say this on Question Time on the BBC, woohoo!
I imagine this is the kind of thing that's happening now with petty crime reports.
Claims about certain categories of crime rising or falling in England are usually based on the Crime Survey for England and Wales, which is not based on police reports, but on surveying a random sample of people to see if they have been self-reported victims of various kinds of crime.
If you don't want your child watching specific shows despite an appropriate age rating, have you considered only letting them watch it while you're with them?
When my child was three, he really liked to watch 'spidey and his amazing friends'. But unfortunately, when he watched it he would emulate some of the bad behaviors from the show, pretend to be one of the bad guys and act out. Easy solution right, we just won't watch the show anymore, we don't leave him alone to watch TV by himself anyways.
Well, on Disney plus, you can't simply hide the show. Even if you remove it from your "recently watched" or whatever, it will show up in preview cards and search results and I'm categories. It became a big friction point, whatever he would see it he would want to watch it. And when Grandma would come over and babysit him, he would ask for it and she'd put it on for him despite our wishes.
So, since then, I've spun up a jellyfin server and ditched Disney plus. If we don't like a show we just remove it, and then it's simply not an option.
It’s also extremely hard saying no to certain shows to my kids, and it would be much easier to just not have them there.
I’m pretty sure the politically oriented people at Disney want this to your kids watch as much of the content as possible, and especially the new ones.
Although it's a lot more effort, if you care a lot about specific things being shown to your children, you could set up your own media server.
You could digitize an existing BluRay or DVD collection and allow your kids to view films and TV using a streaming service-like interface. These days most of the solutions don't even require you to transcode the films, you just RIP them to an ISO and put them on an accessible Samba share and as long as you rename the files to something approximate to the title of the film it'll fetch the metadata for you.
It's even worse if your subscription is packaged with Hulu. Then all the Hulu kids stuff gets pulled in automatically, which includes all kinds of garbage that I very much do not want my daughter watching. All the YouTube-based influencer kid shows with little kids who show off new toys and extravagant vacations each episode.
My only solution was to unsubscribe from Hulu, which stinks because they do have some good stuff.
It really is the epitome of enshitification considering how easy it would be to implement a block button at the very least.
It was extremely good at the other side too. You just have to ask. I can imagine most people don't try this, but LLMs literally just do what you ask them to. And they're extremely good and weighing both sides if that's what you specifically want.
So who's fault is it if you only ask for one side, or if the LLM is too sycophantic? I'm not sure it's the LLMs fault actually.
reply