I do not imply to be dramatic…
Buttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttt
AI video and picture turbines like Sora, Runway, and Midjourney have an enormous downside that might be a deadly flaw.
AI-powered picture and video turbines flip textual content into visuals with what seems like digital wizardry. You sort in a immediate, the algorithm processes it, and a totally rendered picture or video seems inside seconds.
It sounds just like the artistic utopia we have been promised, however is it?
This “utopia” has a darkish facet you would not consider; one may argue it is really radioactive.
And I do not find out about you, however I choose my radios not lively (Ba dum tss).
On this article, I’ll share with you:
- What text-to-video is and why it issues
- What this deadly flaw is and the way we obtained right here
- My genius thought to repair the business
So seize a espresso and kick up your toes. We’re about to leap off the highest rope with this scorching take.
Fast intro
In case you’re new to my work, my title is Lester, however be happy to name me Les
I am a founder with a profitable exit below my belt. As of late, I am the exec chair for a bunch of e-commerce manufacturers, however at my core, I am an award-winning efficiency marketer.
Information and insights are my experience, and my expertise as a enterprise skilled and creator provides me a singular perspective on the vital flaws plaguing the AI picture and video technology business.
In case you’re into data-driven enterprise and advertising insights and methods, try my free publication, No Fluff Simply Info. I share what’s working on the earth of enterprise, together with advertising insights and developments I am seeing.
However that is sufficient about me. I have to carry you up to the mark on what’s taking place with the AI textual content and video technology business.
AI textual content and video technology?
Let’s first break down what AI textual content and video technology is and the way it works.
Right here is the oversimplified model: An AI mannequin learns to generate pictures by first taking present pictures and progressively including noise, making them appear like static on a TV.
It then trains by reversing that course of, studying how one can take away noise step-by-step till it may possibly reconstruct a picture from pure randomness.
Whereas studying, the mannequin is given captions that describe every picture, like “a cat taking part in,” serving to it affiliate textual content with visible patterns. As soon as educated, should you sort new phrases like “a pink automotive,” the mannequin begins with a loud picture and refines it into one thing that matches the outline.
The identical course of occurs for movies however includes making a sequence of pictures whereas making certain clean transitions between frames. The mannequin cleans up noise and learns to keep up movement consistency in order that objects do not randomly change form or place between frames.
Like I stated, that is the oversimplified model. The “what” these fashions do is straightforward; the “how” they do it’s extremely complicated.
This know-how is highly effective however raises vital questions on the way it’s educated and who advantages from its creations.
The massive downside with AI textual content and video turbines
So, what is the huge downside? AI fashions practice on present content material, however whether or not that constitutes truthful use or one thing extra problematic continues to be up for debate.
“Coaching” occurs on the backs of creators who, generally, weren’t requested for consent or given compensation. If you hear that these AI instruments are “creating,” that is considerably deceptive. One may argue they’re compiling, not creating.
As a result of the content material was by no means given to the AI for coaching, it opens a can of worms for copyright claims and questions on who owns the mental property.
Here is the place it will get difficult: Our authorized programs have not caught up but, so companies are extremely cautious about utilizing AI-generated pictures and video.
At ZDNET, we won’t use AI to create content material or AI pictures #jobsecurity.
Moreover, ChatGPT may by no means be as humorous as me, pffft, however I digress.
On Feb.11, 2025, the U.S. District Court docket for the District of Delaware dominated in favor of Thomson Reuters within the case of Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc. The courtroom discovered that Ross Intelligence’s use of copyrighted materials to coach its AI system was not truthful use.
This landmark determination carries vital implications for AI copyright instances. Like I stated, RADIO!!! ACTIVE!!!
My huge thought to avoid wasting the AI business
Now that you’re all caught up, let me let you know about my genius thought.
What if… hear me out: what if we made Spotify however for AI textual content and video? Let’s name it a cool title like AI-text-ify.
AItextify would function equally to Spotify, the place a creator (let’s name him John) uploads authentic work to coach the AI fashions. When a consumer purchases AI-generated content material that makes use of John’s work to create the ultimate output, John will likely be compensated for his work, much like how artists receives a commission for streams on Spotify.
AItextify stands out by making certain creators obtain ongoing funds every time their work is used, reasonably than a one-time payment for coaching. This ensures steady revenue as their content material helps refine and enhance the fashions.
AItextify is groundbreaking as a result of it makes use of blockchain know-how to ensure clear monitoring and truthful compensation for creators. Each time its AI fashions use a creator’s work, they receives a commission immediately and securely.
This enterprise mannequin attracts top-tier creators by providing truthful compensation and elevates AItextify as finest at school.
AItextify additionally features a suite of AI-assisted instruments to assist creators refine their concepts and generate new content material based mostly on developments and consumer preferences. This enables customers to collaborate with AI to create distinctive tales, movies, or artwork that resonates with audiences worldwide.
To make it much more accessible, AItextify will companion with main instructional establishments to supply programs on AI content material creation. This manner, creators from all backgrounds can be taught to leverage AItextify’s instruments to supply high-quality content material.
However why cease there?
AItextify can patrol the web for copyright infringements on John’s behalf. This protects creators’ rights and ensures company shoppers can use AItextify with out worrying about authorized points. That is essential for skilled customers who want dependable and legally sound outputs.
Slap a .AI area on it, and unicorn standing is sort of assured. (As I write this, I am attempting to determine what coloration Lambo to get.)
AItextify can be a win-win-win: creators obtain their justifiable share of the income, customers take pleasure in unparalleled AI-generated content material, and company shoppers can use this content material with out consistently wanting over their shoulders and worrying about potential lawsuits.
My two cents
I believe we simply saved the AI business. Who knew that merely paying creators pretty might be a revolutionary enterprise mannequin?
May this work? In fact, it may; it is a good thought! But when I have to be vital, this is what may go fallacious, except for the whole lot…
I do know related concepts exist, however my concern is that these platforms deal with creators as an afterthought reasonably than an integral a part of the undertaking.
Trade leaders have to prioritize valuing creators. Doing what is true shouldn’t be up for debate. As AI continues to evolve, it is going to be fascinating to see how issues unfold.
If the Thomson Reuters case is any indication, we could also be shifting towards a fairer system for all events concerned.
In case you determine to carry AItextify to life, you will have a CMO. I volunteer as tribute.
Hope this helps; I am rooting for you.
BTW, should you’re into data-driven enterprise and advertising insights, it is best to join my free publication, No Fluff Simply Info. I share what’s working, the newest developments, revenue-scaling methods, and actionable ideas.
Plus, you may get the occasional pep speak to maintain you impressed. If this seems like your jam, click on right here to enroll, it’s very FREE!
Oh, I believe I’ll go together with inexperienced for the Lambo with the license plate “LAMBO-IFY”