Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Apple Releases Open Weights Video Model

starflow-v.github.io

Post reply on HN
Print view View original

151–160 of 178 posts

Re: Apple Releases Open Weights Video Model

#151

Post by kulahan »

devinprater wrote: post #147
badmonster wrote: post #17

Earlier quoted context omitted.

What other accessibility features do you wish existed in video AI models? Real-time vs post-processing?

Mainly realtime processing. I play video games, and would love to play something like Legend of Zelda and just have the AI going, then ask it "read the menu options as I move between them," and it would speak each menu option as the cursor moves to it. Or when navigating a 3D environment, ask it to describe the surroundings, then ask it to tell me how to get to a place or object, then it guide me to it. That could be…

Weird question, but have you ever tried text adventures? It seems like it's inherently the ideal option, if you can get your screen reader going.

Re: Apple Releases Open Weights Video Model

#152

Post by asadotzler »

hrimfaxi wrote: post #63
travisjungroth wrote: post #46

Earlier quoted context omitted.

It might be that you’re not perfectly clear on what exactly you’re trying to convey with the image and why it’s there.

What would you put for this? "Graph of All-Transactions House Price Index for the United States 1975-2025"? https://fred.stlouisfed.org/series/USSTHPI

a plaintext table with the actual data

Re: Apple Releases Open Weights Video Model

#153

Post by asadotzler »

andy_ppp wrote: post #104
devinprater wrote: post #6

Apple has a video understanding model too. I can't wait to find out what accessibility stuff they'll do with the models. As a blind person, AI has changed my life.

I wonder if there's anything that can help blind people to navigate the world more easily - I guess in the future AR Glasses won't just be for the sighted but allow people without vision to be helped considerably. It really is both amazing and terrifying the future we're heading towards.

AURA Vision for blind and low vision people has been doing this for years. Be My Eyes has been doing this for years without AI. Meta Ray-Bans can do this. There's nothing new coming soon that hasn't already been available for a while, only refinements.

Re: Apple Releases Open Weights Video Model

#154

Post by Aloisius »

limagnolia wrote: post #144
gorgoiler wrote: post #98

It’s not really relevant to this release specifically but it irks me that, in general, an “open weights model” is like an “open source machine code” version of Microsoft Windows. Yes, I guess I have open access to view the thing I am about to execute! This Apple license is click wrap MIT with the rights, at least, to modify and redistribute the model itself. I suppose I should be grateful for that much openness, at l…

I think you are looking at the code license, not the model license.

No, it's the model license. There's a second license for the code.

Of course, model weights almost certainly are not copyrightable so the license isn't enforceable anyway, at least in the US.

The EU and the UK are a different matter since they have sui generis database rights which seemingly allows individuals to own /dev/random.

Re: Apple Releases Open Weights Video Model

#155

Post by throwup238 »

SV_BubbleTime wrote: post #100
throwup238 wrote: post #53

Earlier quoted context omitted.

You can safely assume that anyone who uses “virtue signaling” unironically has nothing substantive to say.

>[People who call out performative bullshit should be ignored because they’re totally wrong and I totally mean it.] Maybe you’re just being defensive? I’m sure he didn’t mean an attack at you personally.

It’s presumptuous of you to assume I was offended.

Accusing someone of “virtue signaling” is itself virtue signaling, just for a different in-group to use as a thought terminating cliche. It has been for decades. “Performative bullshit” is a great way to put it, just not in the way you intended.

If the OP had a substantive point to make they would have made it instead of using vague ad hominem that’s so 2008 it could be the opening track on a Best of Glenn Beck album (that’s roughly when I remember “virtue signaling” becoming a cliche).

Re: Apple Releases Open Weights Video Model

#156

Post by BeFlatXIII »

Workaccount2 wrote: post #93
javcasas wrote: post #39

Earlier quoted context omitted.

Finally good news about the AI doing something good for the people.

People need to understand that a lot of angst around AI comes from AI enabling people to do things that they formally needed to go through gatekeepers for. The angst is coming from the gatekeepers. AI has been a boon for me and my non-tech job. I can pump out bespoke apps all day without having to get bent on $5000/yr/usr engineering software packages. I have a website for my side business that looks and functions pr…

AI is divine retribution for artists being really annoying on Twitter.

Re: Apple Releases Open Weights Video Model

#157

Post by MyFirstSass »

alwillis wrote: post #145
MyFirstSass wrote: post #102

Earlier quoted context omitted.

I highly doubt "pumping out bespoke apps all day" is possible yet besides 100% boilerplate, and when possible then no good for any other purpose than enshittifiying the web, and at that point not profitable because everyone can do it. I use AI daily as a senior coder for search and docs, and when used for prototyping you still need to be a senior coder to go from say 60% boilerplate to 100% finished app/site/whatever…

> I use AI daily as a senior coder for search and docs, and when used for prototyping you still need to be a senior coder to go from say 60% boilerplate to 100% finished app/site/whatever unless it's incredibly simple. I know you would like to believe that, but with the tools available NOW, that's not necessarily the case. For example, by using the Playwright or Chrome DevTools MCPs, models can see the web app are it…

I still don't believe that. Again yes a boilerplate calculator or recipe app probably, but anything advanced real world with latency issues, scaling, race conditions, css quirks, design weirdness, optimisation - in other words the things that actually require domain knowledge i still don't get much help with, even with Claude Code, pointers yes but they completely fumble actual production code in real world scenarios.

Again it's the last 5% that takes 95% of the time, and those 5% i haven't seen fixed with Claude or Gemini, because it's essentially quirks, browser errors, race conditions, visual alignment, etc etc. All stuff that completely goes way above any LLM's head atm from what i've seen.

They can definitely bullshit a 95% working app though, but that's 95% from being done ;)

Re: Apple Releases Open Weights Video Model

#158

Post by Workaccount2 »

MyFirstSass wrote: post #129
Workaccount2 wrote: post #108

Earlier quoted context omitted.

Often the problem with tech people is they think software only exists for tech or for being sold to others from tech. Nothing I do is in the tech industry. It's all manufacturing and all the software is for in-house processes. Believe it or not, software is useful to everyone and no longer needs to originate from someone who only knows software.

I'm saying you can't do what you're saying without knowing code at the moment. You didn't give any examples of the valuable bespoke apps that you are creating by the hour. I simply don't believe you, and the arrogant salesy tone doesn't help.

LLMs can pretty reliably write 5-7k LOC.

If your needs fit in a program that size, you are pretty much good to go.

It will not rewrite PCB_CAD 2025, but it will happily create a PCB hole alignment and conversion app, eliminated the need for the full PCB_CAD software if all you need is that one toolset from it.

Very, very, few pieces of software need to be full package enterprise productivity suites. If you just make photos black and white and resize them, you don't need Photoshop to do it. Or even ms paint. Any LLM will make a simple free program with no ads to do it. Average people generally do very simple dumb stuff with the expensive software they buy.

Re: Apple Releases Open Weights Video Model

#159

Post by satvikpendem »

gorgoiler wrote: post #98

It’s not really relevant to this release specifically but it irks me that, in general, an “open weights model” is like an “open source machine code” version of Microsoft Windows. Yes, I guess I have open access to view the thing I am about to execute! This Apple license is click wrap MIT with the rights, at least, to modify and redistribute the model itself. I suppose I should be grateful for that much openness, at l…

> Yes, I guess I have open access to view the thing I am about to execute!

Better to execute locally than to execute remotely where you can't change or modify any part of the model though. Open weights at least mean you can retrain or distill it, which is not analogous to a compiled executable that you can't (generally) modify.

Re: Apple Releases Open Weights Video Model

#160

Post by gaigalas »

givinguflac wrote: post #61
fragmede wrote: post #37

Earlier quoted context omitted.

Turn on advanced data protection so they don't train on yours.

That has nothing to do with it, and Apple wouldn’t train on user content, they’re not Google. If they ever did there would be opt in at best. There’s a reason they’re walking and observing, not running and trying to be the forefront cloud AI leader, like some others.

Why should I buy this "ethical Apple" argument?

They shared audio Siri recordings with contractors in 2019. It became opt-in only after backlash, similar to other privacy controversies.

This shows that they clearly prioritize not being sued or caught, which is slightly different from prioritizing user choices.

Post reply on HN