Whoever is taking shorts from actual popular creators and putting some kind of other content beside it and making a new short is raking in the $ if my feed has been any indicator lately.
Ooh, I'm gonna take someone's 'hot take' reel, superimpose my useless mug over top, and communicate EXCLUSIVELY with agreeable nods and eyebrow raises thereby adding nothing while driving views to my own content instead..
Seriously, that's what it is now. Take someone else's stuff, superimpose yourself silently reacting to it, post. That and spiraling into misery.
Ooh, I'm gonna take someone's 'hot take' reel, superimpose my useless mug over top, and communicate EXCLUSIVELY with agreeable nods and eyebrow raises thereby adding nothing while driving views to my own content instead..
Seriously, that's what it is now. Take someone else's stuff, superimpose yourself silently reacting to it, post. That and spiraling into misery.
I mean even that is at least SOMETHING, a reaction has some content no matter how minimal.
I mean ones like this where it's something totally different on one side.. some video of video games, or machining, or food prep.. terrible.
__________________ Uncertainty is an uncomfortable position.
But certainty is an absurd one.
Knowing that you're one of the people responsible for filling what little social media I do use with AI-generated slop kinda makes me want to hoof you in the groin.
I've seen his slop... it's significantly better than the other slop. And (without giving his idea away), it's clever.
Fuzz will get a kick from this one especially considering its pretty nefarious misinformation use.
Now this was done with Veo 3 before image to video was possible with it, and before you could get good character consistency (plain text to video would get a good likeness but that's it), and honestly the article is already old news.
Earlier this week Google Veo 3 just added the ability to do image to video with sound, and quite frankly unless you knew the persons voice if using a real photo, you will have extreme difficulty ever realizing this is a deepfake, with much better control on how you want the video to look. While image to video certainly isn't new (and neither is lipsync), what is remarkable is just how good and realistic the voice and lipsync is with Veo 3 on image to video using only the image and a prompt. One big drawback so far is it doesn't do vertical videos yet but that is likely to change in the near future.
I'm being paid to promote it this week with free access for promotion purposes and it's really something else (you can use it with Gemini Pro / Ultra as well and Pro often has the first month free if you want to try it out).
This is perfect if you want to create your own persona, or want to generate videos with voice acting with a consistent style (for a movie for example).
"Nayyar said he was trying to make something that looked as realistic as possible, but at the same time he claims people would use "common sense" and conclude they were made with AI."
Hahah, sure they would. I can't believe this guy actually spoke to the media.
He's running into a common failure with satire... if it's not absurd enough, it's not clear in the slightest that it's a joke. A white guy complaining about immigrants on the internet is not exactly the stretch this guy thinks it is (realistically, he 100% knew exactly what he was doing).
It's interesting... everybody was worried we'd see politicians using AI video to trick people, but it's more small-time hucksters doing it. We're going to have to train people up fast to try and protect themselves from nefarious videos online. It's gonna get real bad if we're not careful (and we're not).
You can tell he's fake because he's holding on to a Tim's receipt for a coffee and doesn't have a job.
But ya, the future is going to be a very uncertain place stuffed with fake BS that is impossible to tell from reality. How do you feel wow'd or amazed by anything if anything can be created? News events will be unreliable and twisted to suit any narrative, with AI videos to prove it. Trust will be a thing of the past. How do legislators make decisions on events without fact of reality? Every country negotiating will have a different reality.
Hell, with people wearing AR glasses in the future that can alter the reality you see, perhaps without you even knowing. Many people could see different events happening, shaped by whatever app you gave permissions to. You literally will not be able to believe your eyes. Even more so with a direct brain interface.
Something about the fact that it's a brown dude making a fake video of a white dude complaining he can't get a job at Tim Hortons because of a Punjabi language requirement is near-unjustifiably hilarious to me.
__________________
-James
GO FLAMES GO.
Quote:
Originally Posted by Azure
Typical dumb take.
Last edited by TorqueDog; 07-13-2025 at 02:18 PM.
The Following User Says Thank You to TorqueDog For This Useful Post:
Cool podcast episode on Tetragrammaton (Rick Rubin)
"Episode Description
Jack Clark is the co-founder of Anthropic, an AI research company focused on building reliable and interpretable artificial intelligence systems. Before Anthropic, he was a policy director at OpenAI, where he shaped strategy, communications, and policy. With a background that spans journalism at Bloomberg, policy leadership, and deep involvement in AI governance, Clark has built a reputation for his expertise in AI safety, co-authoring influential research papers and launching the widely-read Import AI newsletter. At Anthropic, Clark guides the company’s work on creating safer and more understandable AI systems, while also engaging in discussions around AI ethics and regulation."
The highly anticipated GPT-5 is right around the corner and its reveal is later this morning for anyone looking to watch the stream. Sam Altman is being a bit sneaky on it, OpenAI has seen its competitors catch up and in some cases even go past them (Google has really stepped up of late). Hyped up products like Sora were huge disappointments and yesterday's news. Many of their employees have been poached, and the foundation behind the company is mainly all gone now. Will it live up to the hype? I highly doubt it. I expect to be quite whelmed but let's see what it can do.
Personal update: I made 650 USD in the past week in paid sponsorships alone. I also got invited by TikTok into a new program where I get paid if brands use my videos for ads (up to 3K per video). I was recruited into the Capcut Creator Program. I got onboarded for ongoing work with one major brand for promoting other products (one I mentioned in a past post). I was invited and met for a chat with the CEO for another popular brand yesterday. While I was paid for my time, the opportunity of meeting the CEO directly was invaluable and will see what comes of this.
More updates to come but will mention them when another major milestone happen.
The AI journey continues...
Last edited by Firebot; 08-07-2025 at 09:19 AM.
The Following 2 Users Say Thank You to Firebot For This Useful Post:
Watching the livestream highlights, some key takeaways: GPT-5 Hallucination rate has been dramatically improved on and was one of the key areas of focus. Wording is more fluid and 'human' and personal like, with a strong removal of words that have a tendency to show it was AI written (words like testament, tapestry, etc).
Coding demo wasn't anything to write home about after using Claude Sonnet 4.0 and Opus 4.1 through Claude Code or even the regular browser
Using this prompt in Claude I got a better one shot game version with Sonnet 4.0 (let alone Opus)
Prompt: Create a single-page app in a single HTML file with the following requirements:
- Name: Jumping Ball Runner
- Goal: Jump over obstacles to survive as long as possible.
- Features: Increasing speed, high score tracking, retry button, and funny sounds for actions and events.
- The UI should be colorful, with parallax scrolling backgrounds.
- The characters should look cartoonish and be fun to watch.
- The game should be enjoyable for everyone.
For fun I created this artifact if you want to see a one shot game with the same prompt with Sonnet 4.0 for apples to apples comparison.
Has anyone had any success with AI CAD yet? I have something I would like to invent but I have very little cad experience, not enough to create my thingamachig anyways. Was wondering if I would be able to have an AI put out the bones for me to tweak a bit before 3d printing something.
I know this is not super relevant stuff, but it would be fascinating to know why they have so much trouble with this sort of thing. I presume it comes down to tokenization of the word, without dissecting the word itself. Once tokenized, it can tell you the colour, shape, properties, everything about it, but the word itself is a single object.
It's also interesting that adding parameters and processing doesn't seem to eventually resolve this. And I wonder if that's just an inherent unsolvable problem given their design. Relating this Tesla to FSD, it may also show that no matter the processing power, it will still phantom brake for shadows and turn into oncoming traffic.
EDIT:
The conversation makes for a fascinating read. Would you trust this type of system to drive your car?
I know this is not super relevant stuff, but it would be fascinating to know why they have so much trouble with this sort of thing. I presume it comes down to tokenization of the word, without dissecting the word itself. Once tokenized, it can tell you the colour, shape, properties, everything about it, but the word itself is a single object.
You were on the right track though the bold is not quite right either. This explains it as well as can be (and tokenization doesn't necessarily happen at the word level but can also be a group of letters, and special characters are their own token).
A helpful rule of thumb is that one token generally corresponds to ~4 characters of text for common English text. This translates to roughly ¾ of a word (so 100 tokens ~= 75 words).
You can test it out yourself here
Last edited by Firebot; 08-07-2025 at 11:26 PM.
The Following User Says Thank You to Firebot For This Useful Post:
Video at the link. Just a reminder not to give your AI bots access to everything. Amazing all you have to do is know they've linked the account, and share a document with them. They don't even need to open or accept the document.
Back in the day we used to share games or other fun programs around that had back doors so you could control someone elses computer. I could see stuff like this being used. "just save the file to your Google drive, then give Chat GPT access, and ..."
Interesting to see OpenAI learn that humans are strange. When an errant update a few months back made ChatGPT openly try to deify the user, they've now dialled that way back and SURPRISE! Now the people that liked the sycophant style are furious.
They're in such an interesting situation. They've been openly mocked for having so many terribly named models that you could select in the dropdown, so they solved it with GPT5 which decides what to use for you... but people also like control and those who learned how to use those terribly named models are freaking out.
I'm more than a little surprised they didn't directly train it on the strawberry question. It's a non-issue unless your job is to legitimately count letters in words, which would be... weird. But it also stands as this direct method for people to lose the plot. Maybe they just don't care about such attacks?
I haven't seen the GPT5 issues many are upset about. It's been quite a bit more to-the-point and I haven't caught it in a hallucination yet. I just had it grab me the emails of 10 people I'm supposed to reach out to today, and not only did it nail 9/10, but the one it missed, it flagged for me that it was taking a stab at it based on the company's general naming convention and suggested I dig further myself on that one.
So just by virtue of hallucinating less and maybe being able to question if it knows something for sure, it's a huge upgrade.
4o was touted as an incredible creative writing model and a lot of people who use it seem upset... but the tests I've run are actually really good. Maybe I'm just blinded by optimism, but I've had it make a few things and they've all been as good or better than 4o.
The Following User Says Thank You to Russic For This Useful Post:
