Millions of people are using (and paying for) LLMs for their daily work. The number of people using crypto as an actual currency is a rounding error by comparison.
There's definitely similarities when it comes to the wave of hype and greed behind them both, but the fundamentals really are completely different.
I work at a company with hundreds of thousands of employees, and they're mandating the use of AI, monitoring it, and pushing like crazy. Like their life depends on it. You get threatening emails if several days pass without you using AI.
Now tell me again what the usage numbers mean in resepect to usefulness.
This is a huge red flag imo. Mandated usage and yet nothing to show for it.
To top it off Sam Altman is a known dishonest actor (and has already shown his true colors at openai). AI is here to stay and has some truly cool uses. But there’s far too many snake oil salesman involved currently.
I don't think it's literally "winner takes all" - I regularly cycle between Gemini, DeepSeek and Claude for coding tasks. I'm sure any GPT model would be fine too, and I could even fall back to Qwen in a pinch (exactly what I did when I was in China recently with no ability to access foreign servers).
Claude does have a slight edge in quality (which is why it's my default) but infrastructure/cost/speed are all relevant too. Different providers may focus on one at the expense of the others.
One interesting scenario where we could end up is using large hosted models for planning/logic, and handing off to local models for execution.
When coming from TS to Dart, the lack of union types was a huge bummer. But I saw a discussion somewhere that you can achieve something close with sealed classes to model union types combined with exhaustive pattern matching, but Idk.
I'm a heavy user of exhaustive pattern matching with sealed classes, definitely a great feature. You can achieve the same outcome as union types, but it's a lot more leg work, e.g.:
```
sealed class MyParam {}
class Bar extends MyParam { String val; }
class Baz extends MyParam { int val; }
void foo(MyParam param) {
switch(myParam) {
case Baz(val: final val):
case Bar(val: final val):
}
}
```
compared to:
```
void foo(Bar | Baz myVar) {
switch(myVar) {
case Bar:
case Baz:
}
}
```
I hadn't seen the extension_type_unions package, though, I'll check it out.
Note for those not clicking, the failure had absolutely nothing to do with resistance from Iran, or anything Iran did. It had to do with operational issues, including 3 failed helicopters out of the 8. Upon Carter accepting recommendations to abort, a 4th crashed resulting in the 8 soldiers lost.
This happened 45 years ago, with different gear, and a lot of operations between now and then. It should have zero impact on any decision to go into Iran or not.
Of course, whether the US would go in is still a valid question, including other aspects of risk.
Indeed the failed Iran hostage rescue led directly to the formation of special operations specific helicopter units and training doctrine we have today.
And it's a really, really, really sore subject in Iran so even if it "goes well" it represents a much larger escalation than doing the same thing in a different country would.
Well, "special operations" that involves foreigners running around in sovereign countries setting of bombs tend to be a sore subject in most countries I can think of, not sure Iran is very special in that regard.
>Well, "special operations" that involves foreigners running around in sovereign countries setting of bombs tend to be a sore subject in most countries I can think of, not sure Iran is very special in that regard.
The entire point of making the comparison to "doing the same thing in another country" rather than nothing or some alternative course of action was to head off this specific nitpick.
The US already tried that thing once in the early days of the current regime (literally linked in the comment I initially replied to!) and it's one of the bigger reasons we don't have normalized relations with them, though it's perhaps a distant second to the 800lb gorilla with a little hat in the region.
Wasn't the operation planned, to rescue hostages that the regime wasn't capable of protecting? In fact, didn't care to protect? Wikipedia seems to say so.
After all, you're supposed to police your own residents, and keep them from invading embassies, which are foreign turf. You're also supposed to respect that aspect of diplomacy. Letting your citizens raid an embassy and kidnap diplomats is the act of a banana republic, and definitely signals the end of diplomacy.
If the current regime is upset at someone trying to rescue their own people kidnapped from their embassy, then frankly that regime is insane. Iran is 100% at fault for allowing that to happen, not working quickly to resolve it. The US is 100% correct to have sent people in to rescue their own people, under those circumstances.
If the embassy was better armed, with more security, they'd be completely correct to shoot-to-kill every single person who stormed that embassy. Yet sending people in to rescue them after the fact is... wrong?
Hardly, and I sincerely doubt Iran is significantly upset still. Yelling about it, sure. Upset for real? No.
So does bombing them from the air and assassinating their senior leadership (and a number of civilians who had the misfortune to be standing near them at the time)!
I'm not sure, the last time the US did this it was in Pakistan and resulted in no escalation consequences whatsoever.
(mind you, the idea that a raid on a fortified facility, rather than lightly or undefended civilian buildings, would be an easy win, is the real delusion)
If I see any story about OpenAI, I assume it's an intentional 'leak'. I know all companies spend money on paid PR but OpenAI (and Sam Altman) take it to a completely different level.
I suspect that generating textures directly in UV space is a bit of a dead end because UV maps are object-specific, so flux/SD/etc are going to have a hard time understanding the inputs.
I've been experimenting with image-to-image (and video-to-video) for basic texture projection, which I think shows promise:
It actually performs remarkably well with mask maps as Control Nets. You can use a simple black background and white UV shells, that will give you OK results. But then you expand and use curvature maps just like you would with line art control nets, and then you have a controlled painting.
I'm a happy Supabase customer. It's not just postgres, it's postgres + auth + functions + storage all in one place, with a very good local Docker environment and a variety of nice client SDKs.
It's a quick/convenient way to get a "full" backend up and running. The other option (Firebase) can't be self-hosted and has some absurd pricing footguns.
No idea at all. My blog does really well on Google, so I think if either site is being penalized for duplicate content it's probably the Substack.
It looks like 8% of my newsletter signups are "from the app" according to the Substack dashboard - which I think is how they show signups that they've encouraged as opposed to signups I had myself.
I'm really just using Substack because they've solved email deliverability and they're free for me to use to send out emails.
If you have a blog with RSS and just need something that auto-emails people your updates you can run my project on a raspberry pi. No need to tie yourself into the substack ecosystem if you don't want to
There's definitely similarities when it comes to the wave of hype and greed behind them both, but the fundamentals really are completely different.