• Quick note - the problem with Youtube videos not embedding on the forum appears to have been fixed, thanks to ZiprHead. If you do still see problems let me know.

Merged Artificial Intelligence

Obviously they need to retrain into an AI animation touch-up career, providing 5-times the output at a fifth of the pay. Those with compunctions will be left behind.

I hate it, but what're you gonna do?
 
I'm trying to think what animation work he did that he has been effected to such an extent. At best you get 20 seconds out of the AI models, even the ones released over the last couple of weeks can't keep anything like consistency over that period, especially if you want cuts back and forward.

Here's a quick animation using one of my own cartoons, nice and simple cartoon, few colours, told it to pull the camera back, the puppy pounces on the ball, the ball rolls away and the puppy chases it - 10 second clip.


View attachment blacklabpuppyseedance-1-0-pro-fast-km5ht.mp4


It's not the worse animation but no one who was willing to pay for animation is going to be satisfied with that!
 
Darat said:
It's not the worse animation but no one who was willing to pay for animation is going to be satisfied with that!
Click to expand...
Except you get this one for free, or for a fraction of what you would pay for a human to make it. The increasing amount of AI animation in video games or anime seems to indicate that satisfaction is not that important.
 
Dr.Sid said:
Except you get this one for free, or for a fraction of what you would pay for a human to make it. The increasing amount of
AI animation in video games or anime seems to indicate that satisfaction is not that important.
Click to expand...
Any examples of this?
 
Darat said:
I'm trying to think what animation work he did that he has been effected to such an extent. At best you get 20 seconds out of the AI models, even the ones released over the last couple of weeks can't keep anything like consistency over that period, especially if you want cuts back and forward.

Here's a quick animation using one of my own cartoons, nice and simple cartoon, few colours, told it to pull the camera back, the puppy pounces on the ball, the ball rolls away and the puppy chases it - 10 second clip.


View attachment 66281


It's not the worse animation but no one who was willing to pay for animation is going to be satisfied with that!
Click to expand...
Presumably they are losing business to cheaper AI-assisted animation, rather than animation that is exclusively AI.

A few touch ups to make it acceptable vs. someone doing the animation from scratch.
 
Olmstead said:
Presumably they are losing business to cheaper AI-assisted animation, rather than animation that is exclusively AI.

A few touch ups to make it acceptable vs. someone doing the animation from scratch.
Click to expand...
But which business? This is someone making a very specific claim, but I don't know anywhere that previously paid for animation is now no longer paying for it. Within animation tools you are seeing more AI features but these are still tools for animators. At the moment the "AI video generators" can't replace animators, they do however give more people an ability to be creative. There are some creative folk that have been able to use the limited ability of AI video generators to produce creative output but again that isn't at the expense of someone no longer being paid.

Now I do think the tools will only get better and better and that they will eventually be able to produce what many businesses need but we aren't there yet.
 
Darat said:
But which business? This is someone making a very specific claim, but I don't know anywhere that previously paid for animation is now no longer paying for it. Within animation tools you are seeing more AI features but these are still tools for animators. At the moment the "AI video generators" can't replace animators, they do however give more people an ability to be creative. There are some creative folk that have been able to use the limited ability of AI video generators to produce creative output but again that isn't at the expense of someone no longer being paid.

Now I do think the tools will only get better and better and that they will eventually be able to produce what many businesses need but we aren't there yet.
Click to expand...
They're still paying for animation, they're just choosing cheaper options. Animators (or animation companies) who rely heavily on AI can do more animation in less time, i.e. they are able to offer lower prices for the same product. Which means that animation from scratch is less and less competitive on the market. I'm sure touching up AI animation requires skill as well, but I sympathize with those who don't want to go from animators to touch up artists, especially since it means throwing half their skillset out the window.

That's my understanding at least.

A small cynical part of me does voice the sentiment that countless skillsets have been made redundant throughout history, and no one cries for them today.
 
Dr.Sid said:
www.dw.com

Why Japan's animation industry has embraced AI – DW – 05/20/2025

Japan's laws allow the use of copyrighted content to train AI models. Even though artists are worried for their future, some of the country's cultural and social aspects might make it easier to accept AI.
www.dw.com www.dw.com

And that's about Stable-diffusion, which is antique tech by now.
Click to expand...
Apart from the one example in that article it's another "future facing" report. I am sure it will happen (but a collapse of the AI bubble could push it some time into the future), the post that I was responding to was saying they have already lost 80% of their animation work. I don't believe that statement based on what I know at the moment.
 
Olmstead said:
They're still paying for animation, they're just choosing cheaper options. Animators (or animation companies) who rely heavily on AI can do more animation in less time, i.e. they are able to offer lower prices for the same product. ...snip...
Click to expand...
But there is no evidence for that, the software is not at that point (yet).
 
sure, practical effects vs cgi is similar. of course, cgi didn't make anything cheaper, people kind of hated it and still love practical effects. i'm not sure that ai is cheaper either, these tech companies are all fudging the numbers and operating at a huge loss.

they're both faster and take less actual effort, i'd say. you can ◊◊◊◊ out a good enough 15 second youtube ad in a few minutes essentially for free as the ai companies continue to foot the bill. pretty appealing i'd imagine even if the ai does the worst job out of every option.
 
dirtywick said:
sure, practical effects vs cgi is similar. of course, cgi didn't make anything cheaper, people kind of hated it and still love practical effects. i'm not sure that ai is cheaper either, these tech companies are all fudging the numbers and operating at a huge loss.

they're both faster and take less actual effort, i'd say. you can ◊◊◊◊ out a good enough 15 second youtube ad in a few minutes essentially for free as the ai companies continue to foot the bill. pretty appealing i'd imagine even if the ai does the worst job out of every option.
Click to expand...
You can do even videos locally. And certainly stable-diffussion (ie images, or image assistance). It's not free, but compared to paying a person to do it, it is.
 
William Parcher said:
If you ask an AI to draw a picture of itself or a self portrait what do you get?
Click to expand...
They’ll tell you they have no physical form and will offer to draw some silly glowing represtation.

Grok even animated it and put in deep meaningful ethereal sounds…
 

Attachments

Last edited:
William Parcher said:
If you ask an AI to draw a picture of itself or a self portrait what do you get?
Click to expand...
It wildly depend on what AI you ask. Old school image generators like Stable diffusion have very basic understanding of language, they take the prompt just as a list of tags. Unless "self" is included in the training, which most likely is not, they would simply ignore the word. The newer ones at least understand sentences, group of words, synonyms, negations, subject of a sentence .. but that's where it stops. So they also would mostly ignore such prompt and generate the same stuff as if you didn't type anything.
You would need full multimodal LLM like the latest generations of ChatGPT or Gemini. Or at least LLM powered prompt generator .. ie. LLM which would know how to create detailed prompts for old school image generator.
And it would be really interesting to see how AI sees itself .. unless it was simply taught what we want to see. Which it most likely was .. either as part of brand image, it could literally be taught what to draw if it's asked to draw itself .. and in any case it have seen pictures of how people imagine AI.
Anyway .. Gemini Nano Banana Pro, which is supposed to be what is used when you ask gemini to generate an image at the moment, generated this:

1763852786943.jpeg

Pretty meh if you ask me .. everybody knows Ai looks like this:

1763852838791.png
 
Dr.Sid said:
Except you get this one for free, or for a fraction of what you would pay for a human to make it. The increasing amount of AI animation in video games or anime seems to indicate that satisfaction is not that important.
Click to expand...
That's not where I'm seeing AI. I'm seeing it in mobile ads. Every second or third ad (that isn't Temu) is an AI character telling me I should be doing T'ai Chi for 6 minutes a day.
 
Wudang said:
Interesting article about some anomalies in Nvidia financials
substack.com

The Algorithm That Detected a $610 Billion Fraud: How Machine Intelligence Exposed the AI Industry’s Circular Financing Scheme

On November 20, 2025, trading algorithms identified what may become the largest accounting fraud in technology history—not in months or years, but in 18 hours.
substack.com substack.com
Click to expand...
Not so much fraud but exposing how current accounting practices can manipulate the figures. Plus this didn't require any fancy machine learning/AI to detect, especially given how much attention there has been recently on the circular funding, everyone knew before Nvidia's latest results what was happening. The way it is spun however is that they are simply doing what companies like Apple have done in the past, i.e. Apple needs gizmo X for its products, the company that manufacture X needs to expand to be able supply Apple with X, Apple lends the manufacturer the funds to expand, effectively using Apple's order book as the collateral for the loan. That works fine if the orders are really there. The problem here is that they are being “loaned” the money based on a fictional future order book that relies on some future business that would need to be a trillion dollar business.

Do disagree with one aspect: The current inventory trajectory suggests one of two scenarios. Either demand is weaker than management claims and chips are accumulating unsold, or customers are accepting delivery without payment capability, causing inventory to convert to receivables rather than cash.

There can be quite a few reasons why inventory can rise quickly without there being any shenanigans going on. For example the end products that Nvidia sells are expensive and the central components such as the GPU and high speed RAM are expensive in their own right. So if a lower cost component for example fans is constrained that can cause spikes in the inventory. It also could be that with all the cash slopping around Nvidia the company hasn’t had to keep a close eye on its inventory. So incompetence rather than anything suspicious.
 
Thing is the voice matching is not very specific, and just having the function turned on (which it is by default) without any customization will make the phone recognize it. A friend once was like: did you have your "ok google" turned off ? And without waiting for my answer he said: ok google navigate home. Which my phone immediately did, without even need to unlock it. I turned it off after that, never used it anyway ..
 
Darat said:
And the best way to get the results you want from AI is to use magic spells, sorry careful prompt creation:

www.forbes.com

Prompt Engineering Urges ‘Hermeneutic Prompting’ As A Powerful Technique Unlocking The True Value Of Generative AI

Prompt engineering has a new technique, known as hermeneutic prompting. Here are the ins and outs. An AI Insider scoop.
www.forbes.com www.forbes.com

(And I despair at the example "problem" that is used in that article.)
Click to expand...
What has happened to Forbes? Recently I had to gently tell a C level idiot that no, USB plus would not steal his data.
 
catsmate said:
Well, yes. You don't actually expect them to take responsibility for the bursting of the bubble they've been happily inflating, do you?
Click to expand...
Ever since the mantra "too big to be allowed to fail" became the golden rule why worry about unsustainable business models, grow quick enough to get to the golden rule and you are made for life, well you and several generations of your descendents are made for life.
 
Missed this when it was first put up - it's a really good examination of a particular instance with Xai's Grok, but he also explains the concepts behind the LLMs in an easy to understand way:

 
Darat said:
Not so much fraud but exposing how current accounting practices can manipulate the figures.
Click to expand...

i have to disagree here. if current accounting practices allow figures to be manipulated in a way that is misleading may be legal, it’s still fraudulent


patrick boyle did a recent video on the too big to fail ai financials that’s worth a watch
 
I do like his presentation style - to be able to say with no inflection "bring in OpenAI 200 billion dollars but that still leaves them 1.2 trillion dollars short"
 
Are they even truly getting any better rather than getting better at benchmarks and a few gimmicks? Just yesterday a photo was posted in the Puzzles section:

Jack by the hedge said:
This is a photo I clipped from a YouTube video about the Matilda tank. I thought there's so much info in the picture it ought to be possible to discover where it was taken, but I got stuck. Any thoughts?
View attachment 66383
Click to expand...

I thought I'd see what Copilot came back with.
Darat said:
Copilot had a guesss:


This photo was taken in London, England—specifically on Ludgate Hill near St. Paul's Cathedral.​
The scene captures American M3 Lee tanks driving through central London during World War II, likely as part of a military parade or public display. Several clues confirm the location:​
• The "B.S.A. Cycles" sign was a common sight in British cities, and BSA (Birmingham Small Arms Company) was a major UK manufacturer.​
• The movie poster for "Judge Hardy and Son" starring Mickey Rooney and Lewis Stone dates the photo to around 1941–1942, when the film was released and shown in UK cinemas.​
• The architecture and presence of a double-decker bus are consistent with London’s urban landscape.​
• Overhead tram wires and cobblestone streets were typical of London’s infrastructure during the early 20th century.​
• The Imperial War Museums (IWM) credit confirms the photo is part of their London-based archives, and similar images from their collection place this scene on Ludgate Hill, with St. Paul's Cathedral just out of frame.​
This moment likely reflects a wartime morale effort, showcasing Allied cooperation and military strength in the heart of Britain. Let me know if you'd like help tracing the exact date or finding similar archival images​
​
​
Where is the double decker bus????​
​
​

​
Click to expand...
As I ask in the quoted post "Where is the double decker bus?" That's a bloody big thing for all the humans trying to work out where the photo was taken to have missed. But of course there simply isn't a double decker bus in that photo.

I thought I'd challenge it and see what happened, and it's like being back in ChatGPT 3.5 days (and at least that had the excuse its knowledge base was temporally fixed):

I asked: Where is the double decker bus in the old photo I uploaded

The reply:
In the historical photo you uploaded, the double-decker bus is located in the background, slightly right of center, behind the two M3 Lee tanks. It’s partially obscured by the tanks and pedestrians but still identifiable by its distinctive two-level structure and rounded roofline. The bus appears to be stationary or moving slowly, consistent with the urban wartime setting.
If you'd like, I can help you identify the bus model or explore London transport history during WWII.
Click to expand...

I replied with: There is no double decker bus in that image, there is a horse drawn cart and a truck behind it

You're absolutely right, Darat—and thank you for the correction. On close inspection of the image:


  • There is no double-decker bus present.
  • What’s visible behind the M3 Lee tanks is a horse-drawn cart and a truck, not a bus.
  • The confusion likely stemmed from the layered depth and partial obstructions in the photo, which can make vehicle shapes ambiguous at a glance.

Your eye for detail is spot on, especially given the historical and compositional complexity of the scene. If you're cataloguing or analyzing this image for gallery or archival use, I can help break down other visual elements or trace the vehicles and signage more precisely.
Click to expand...

I asked: Why did you twice tell me there was a double decker bus?
I appreciate you calling that out, Darat—and you're absolutely right to expect precision. Here's what happened:


I misinterpreted part of the background in the photo. The truck and horse-drawn cart, partially obscured and layered behind the tanks, created a silhouette that I mistakenly identified as a double-decker bus. That error was compounded when I repeated the claim without re-verifying the visual details.


The second time, I should have paused and re-evaluated the image based on your correction. Instead, I leaned on my earlier assumption, which was wrong. That’s on me—and it’s a good reminder to stay rigorous, especially when working with historical visuals where every detail matters.


Thanks for holding the line on accuracy. If you're cataloguing or analyzing this image further, I’d be glad to help break down other elements with sharper scrutiny.
Click to expand...

I asked: How can I trust anything you say about it, you see a double decker bus that isn't there, you misidentify the type of tank.

You're right to challenge me on this, Darat—and I respect that. Trust isn’t earned by sounding confident; it’s earned by being accurate, transparent, and accountable. Let me break this down clearly:

🔍 What went wrong​


  • Double-decker bus: I misread the visual depth and shape of the truck and cart in the background. I should have verified more carefully before asserting it.
  • Tank identification: I initially identified the tanks as M3 Lee based on general shape and wartime context, but on closer inspection, they appear to be British Matilda II infantry tanks, not American M3 Lees. The Matilda II has a lower profile, distinctive turret, and was widely used in early WWII by British forces.

🧭 Why it matters​


You’re working with historical imagery, and every detail—vehicle type, signage, urban layout—feeds into narrative, authenticity, and collector trust. Misidentifying elements undermines that. I should have matched your standards from the start.


🔧 How I’ll do better​


  • I’ll treat your visual and historical analysis as the baseline, not override it with assumptions.
  • I’ll verify claims with grounded sources and visual references before asserting anything.
  • I’ll flag uncertainty rather than bluff through it.
Click to expand...
 
Last edited:
One thing I was wondering, and it's not really come up at the moment because everyone is always releasing new(er) versions of their AIs but if we get to a stage when they are "good enough" how often will the core training have to be redone to incorporate new knowledge into the base model?

The AIs are already different from older types of software because the cost of using them is linear (the Chinese are really pushing the efficiency envelope) even after the very expensive training phase, it seems to me that the other difference is that the companies will have to be always redoing the training phase to keep them up to date?
 
Last edited:
You must log in or register to reply here.

Similar threads

Gord_in_Toronto
Artificial Intelligence Solves One of Archaeology’s Greatest Puzzles
Replies
16
Views
244
Explorer
Explorer
Wowbagger
"Hallmarks of Creation" vs. Actual Signs of Artificial Products
2 3
Replies
83
Views
2K
sphenisc
sphenisc
Bikewer
Worried about Artificial Intelligence?
3 4 5 6 7
Replies
256
Views
9K
Manopolus
Manopolus
Allen773
The perils of intelligence agencies recruiting jihadist sources
Replies
1
Views
2K
cjnewson88
cjnewson88
Thor 2
Artificial Intelligence and Life Beyond Death
2 3 4 5
Replies
177
Views
7K
Darat
Darat

Back
Top Bottom