Those birds!!!
Exactly. Itâs the total opposite of what well planned product development and launch would call for. Tease the possible, keep the roll-out contained to the feasible, and slowly build on that once youâve proven that the product stands up to scrutiny.
The problem is this could last another 2, 5, 10 years on the AI video side. And they donât have that much runway.
One thing Iâve always found to be the #1 challenge for freelancers and entrepreneurs - balancing sales and making products (software, hardware, or artistic kind).
Put an engineer into a sales position and it will not go well. The engineer (or artist) is all to familiar with the limitations, the things that havenât been finished, refined, etc. He will be reluctant to promise a prospective customer or even refuse a lead, because he/she thinks thereâs no way the product is ready for what the customer surely will expect.
Put a sales person into the position without restraint, and they will fill the pipeline more than you can handle. A good sales person knows just enough to talk about the product in a believable way and can answer questions about capabilities genuinely, but is also not afraid to push reality, bend truth a bit, and use ignorance to make it through the day. Theyâre story tellers, not makers.
When youâre a freelancer or small business person, you have to wear both hats, and itâs almost impossible to pull off. Thatâs why most self-employed artists are starving or under performing business wise.
In terms of OpenAI and other outfits - people like Sam Altman are sales people, not engineers. No fault there, just keep that in mind. They are very good and silencing the engineers from leaking out the truth about the products and keep the con going. Lots of money and the promise of 47 virgins goes a long way at times.
But thatâs why we have such a negative reaction. We donât look at this like a sales person does, we look at this as the engineer or artist, and just go OMG, who let that person talk publicly. Itâs not ready!!!
Clearly ToysâRâUs was happy with the results or they wouldnât use it on the homepage. At least the ToysâRâUs executives. The branding people might have a few reservations.
What artists/engineers think the customer wants/expects and what the customer actually considers the bar is often different. Not talking Netflix here, but more general terms.
From this morningâs YT feed. A really good perspective on AI.
Listen to Adamâs first question & answer, and then replace Walmart with AI. Rest of the logic totally applies. AI output is the black fence, until it needs to get painted whiteâŚ
Sometimes you need to let the world solve these problems to quote his take away. And maybe thatâs a good perspective for creatives in this overall mess.
Oh, come on! It obviously says âItfallingrami codmoxzzâ
I think they are archeopteryx.
Pterodactyls!
Is that the AI or the VFX version of Lorem Ipsum?
These frames look like a live action Mad Magazine. Itâll work for making unsettling images maybe a horror film or something but will it make people want to buy stuff?
I think the only reason anyone was ok with this was because itâs still worth a few clicks just to say âWe made our new spot entirely with AI!â I mean itâs the only reason any of us watched it.
The problem AI has it that is gonna be click-worthy for maybe a couple of months more. After that itâs going to need to not look like garbage and, as I think you were also saying, they donât have that much time. I mean if any of us had tried delivering this to an agency weâd get crucified.
I got as far as 20 seconds - couldnât face the next seven minutes.
Back to 1990s wallpaper sci-fi with humans, voices and musicâŚ
This afternoonâs treat will be 1930s wallpaper sci-fiâŚ
While technically a bit more cohesive, still just an animated neon slideshow for the most part.
This comment was actually the most meaningful:
Yeah, it looks beautiful and all but still a glorified animated slideshow generator. Still no generation of any sets of meaningful actions.
I want to see dudes coming out of their houses and locking the doors behind them, people feeding birds in parks, a girl playing a full round of hopscotch, someone stopping a taxi and jumping in.
Even something as simple as someone grabbing a bottle of water, opening it, drinking from it and putting it back down, Sora or any video model canât do thatâŚ
Those AIs canât handle any generations that involves anything that requires more that 2 or 3 steps in even minimalistic thought processes and scaling probably wonât fix that since those are things that actually require video generation models to have some kind of foundation that would, if you really think about it, need to be at least as smart as an AGIâŚ
You canât wing it without some kind of agentic and feedback loop driven generation process.
At least it wasnât the Pillsbury Doughboy faceâŚ
âGymnastics is the Turing Test for video generation modelsâ
No notes.
Fix the phone number and itâs perfect.