By Ian Hollidae, 2025/10/07

This is one of the better responses to the latest over-the-top, sensational story about AI generated something I've seen to date:

AI actors are a long con... a grift... another bull**** tech fantasy Silicon Valley is using to drum up another round of sweet, sweet V.C. cash. It's just another iteration of the same insanity which convinced Hollywood to toss out the theatrical film and ad-supported network TV models in favor of streaming-based subscription fees... a decision which almost everyone in the industry now regrets.

-- George MF Washington, regarding the Tilly Norwood buzz

Personally, I see Tilly as nothing more than a logical next step: a fancy, non-cartoonish animation.

As for the larger implications for Hollywood, maybe this is the twenty-first century equivalent of talkies replacing the silent film. Or maybe we're now getting the first full view of a completely democratized media landscape. Regardless of what has been tossed aside, or where things are going, profit motives will always decide everything. Unfortunately, no one knows where the money trail is right now.

By Ian Hollidae, 2025/10/01

So OpenAI has decided that mixing social media and AI (generative video) is somehow a good thing. Yesterday's announcement highlights all the things you can possibly do with their new Sora 2 model and iOS app:

* From words to worlds: Start with a prompt or upload an image to create videos with unprecedented realism in any style: cinematic, animated, photorealistic, or surreal.

* Cast yourself and your friends: Create together. With cameos, you and your friends can be characters in your videos. You control how or when your cameo is used.

* Remix everything: Take someone else's creation and put your spin on it. Swap characters, change the vibe, add new scenes, or extend the story.

* Cue the sound: Music, sound effects, and dialogue are automatically included in videos to make every scene complete.

The most obvious question is how is this better than shooting a real video and utilizing all the same features The next most obvious question is does anyone really need this. The third most obvious question is will this be the "killer app" for generative media (or at least enough of a success for VC's to recoup their investments).

I'm sure people will come up with all sorts of responses. I'm guessing a lot of them will not be overly favorable to OpenAI. This all still feels like scrounging around for questions that no one has asked to fit answers only AI companies are shouting about from rooftops.

By Ian Hollidae, 2025/06/24

So we finally have some sort of legal direction regarding LLM's training on public material.

The tech industry will call this a win. The content industry, not so much. From the court summary (full PDF at the link above):

The copies used to train specific LLMs were justified as a fair use. Every factor but the nature of the copyrighted work favors this result. The technology at issue was among the most transformative many of us will see in our lifetimes.

The copies used to convert purchased print library copies into digital library copies were justified, too, though for a different fair use. The first factor strongly favors this result, and the third favors it, too. The fourth is neutral. Only the second slightly disfavors it. On balance, as the purchased print copy was destroyed and its digital replacement not redistributed, this was a fair use.

The downloaded pirated copies used to build a central library were not justified by a fair use. Every factor points against fair use. Anthropic employees said copies of works (pirated ones, too) would be retained forever for general purpose even after Anthropic determined they would never be used for training LLMs. A separate justification was required for each use. None is even offered here except for Anthropic's pocketbook and convenience.

And, as for any copies made from central library copies but not used for training, this order does not grant summary judgment for Anthropic. On this record in this posture, the central library copies were retained even when no longer serving as sources for training copies, hundreds of engineers could access them to make copies for other uses, and engineers did make other copies. Anthropic has dodged discovery on these points. We cannot determine the right answer concerning such copies because the record is too poorly developed as to them. Anthropic is not entitled to an order blessing all copying that Anthropic has ever made after obtaining the data, to use its words

As someone who spends time on both the tech and the content side of things, I think the content side will eventually come out just fine after this. I'm not sure I could say that before now.

First, it appears that copyright precedence was followed. If you download/obtain something legally, it falls into fair use. This seems pretty straight forward. Some part of what Anthropic did cleared the established legal hurdles.

Secondly, storing copyrighted material in a permanent database without compensating the owner is considered piracy. This part of what Anthropic did fell short of the established legal hurdles and will go to trial later this year. I think it was this aspect of LLM training that really cause a lot of angst. The idea of web crawlers scouring the web without restraint and with no regard for ownership seemed inevitable. I'm glad to see some rational lines being drawn.

In some ways, this ruling is a bit of a relief. Enough of the talk and speculation and the start of rubber-meets-the-road action. Looks like the results are off to a good start.

By Ian Hollidae, 2025/05/23

If you need any more evidence that the AI video generation age is being pushed forward, look no further than the steady stream of announcements. Googles recent event was full of the normal demos and the web was full of the normal end of the world comments.

But if you watch closely, there is something else that seems to be emerging. Has anyone noticed that none of these products promise *better* video/film creation than what's available through traditional means? Was there anything done in "Freelancers" (one of the demos) that can't be done today? For all the technological advances suposedly being made, I would think these announcements would present more than the predictable platitudes of "new chapters in creativity".

Now I get that video AI might settle down as a tool, and these tools need to be in the right hands, but these services don't seem to be offering better video even on a purely technical level. I find this odd mainly because all one has to do is dig up a video from decades ago to see just how much video, technically, has improved without the use of AI. Higher resolution screens, better means of video recording, a whole slew of new codecs, the list goes on. Even in the old days, there was never any question about the quality of video improving moving forward.

I think regardless of what happens next, it's clear the only value video generation has is to cut production costs. And that's only if the model somehow manages to create what you want before reaching your monthly service limit. Don't get me wrong, costs do matter. But if AI isn't really improving creativity, and the cost savings aren't realized under real world situations, then none of this will move forward for very long.

By Ian Hollidae, 2024/11/20

Up until now, I've been hesitent to make a post about AI mainly because it's a topic that's already received plenty of attention. What more is there to say? Sometimes new technology makes a big difference and sometimes it has minimal to no impact. In a lot of cases, new technology is just an evolution of something that exists.

Obviously, AI has been the big rage in tech over the last couple of years. If you listen to most of what's being written, AI threatens to end civilization as we know it. And while I'm not here to dismiss peoples concerns, this is why I decided to try and sprinkle in a few thoughts that are a little less gloom and doom.

Progress Is Progress

Looking back over history, everyone can name a case where technology actually did replace something. The automobile replaced the horse and carriage. The electric lightbulb replaced oil lamps and candles. Internal combustible engines mostly replaced steam engines.

On the flip side, there are plenty of examples where technology didn't replace anything at all. Television didn't replace movies. Microwaves didn't eliminate cooking. Movie theaters weren't replaced by VCRs. It seems to me that the history of new technology falls into one of four categories regardless of what the new invention is:

1. New technology completely replaces XYZ
2. New technology leads the way in a new process with humans working in support
3. Humans lead the way in a new process with new technology working in support
4. Nothing happens at all

Right now, the big issue in media is the AI generation of audio, video and photography. Can it replace professional photographers, videographers and musicians? Things are clearly in the early stages and I think more new questions will arise before a lot of concrete answers appear. However, it's obvious the success of AI will be as simple as answering the basic question of will it be useful. But media generation is a fairly straightforward issue and I don't see much in the way of expansive use. Make a video. Make a photo. Generate some audio.

On the other hand, the use of AI in supporting tasks, activities surrounding media creation, seems like the area that will be impacted most. Since I think some form of AI will probably be with us for the foreseeable future, I thought it might be interesting to share a few items of what I'd like to see. I have to add that the services I reference aren't endorsements. They're just things I'm looking into.

A Few Easy Suggestions

The first item I would like to see is the ability to use AI in predicting weather conditions at a desired shoot location. It would be nice if a service could could access as many weather models available to show all possible conditions. Making predictions as far out as possible would be even more helpful. Right now, it seems the most popular method of seeing weather conditions is to look at a bunch of different weather apps and guestimate. Services such as Atmo and Climavision already seem to be laying claim to this space.

Second, using AI to find "off the beaten path" shoot locations. Similar to my weather request, an AI service that's a combination of 3D Google Earth and a popular sun tracker to find shoot spots could attract a nice audience. Location and tracking apps already exist but I'm thinking AI could provide far more advanced visual and data presentation features.

Thirdly, I'd like to see a dedicated AI texture generator. As someone who does X3D, finding or creating the right textures can take quite a bit of time. With AI, just input a text prompt, get back a texture. All the current leaders in imaging, namely Adobe, seem to be focused more on traditional-ish photo creation. I'm sure I could use their AI tool for textures but I'm thinking about a far more task related service. Services such as OpenArt and Polycam look to have a headstart.

I Think I'll Stop Here

Obviously, I'm sure there are plenty of other ways AI can be utilized without replacing everyone.

Anyway, who knows how this will all finally turn out. One thing for sure is that actual results will eventually replace talk and I'll let this be my brief dive into the AI discussion.

For now.