- Smoking Robot
- ⚖️ US Copyright Office Deals Devastating Blow To AI Art
⚖️ US Copyright Office Deals Devastating Blow To AI Art
Why AI images can't be copyrighted, researcher explains in great detail how AI will kill us all, Headlines, Tool of the Day, Links and more
Good morning, SmokeBoters.
Welcome to our new subscribers, including some of you BIG DAWG media types, VCs, and the random guy with a Duck Duck Go email address who looks like someone super famous mentioned in this email.
In the email today:
US Copyright Office says AI images can’t be copyrighted ⚖️
AI researcher predicts super intelligence will kill us all ☠️
Tool of the Day 🔨
LET’S GET DARK!
US Copyright Office Says AI Images Can’t Be Copyrighted ⚖️
In a previous send, we detailed how the intellectual property lawsuits related to AI are apt to clog court dockets for years.
Well, here comes the US Copyright Office with the firmest ruling thus far on what you can and can not copyright in AI.
Our story begins with graphic novelist Kris Kashtanova, who recently produced “Zarya of the Dawn”, a comic book with AI-generated images for which she sought a copyright.
The Copyright Office granted her IP protection on the text and arrangement of the images. But it denied IP protection for the AI-generated images.
Here’s a super nerdy deep dive video I did on it:
The U.S. Copyright Office has ruled that images created by Midjourney in the graphic novel "Zarya of the Dawn" should not be granted copyright protection.
The decision means author Kris Kashtanova is entitled to copyright for the text and arrangement of the images, but not for the images produced by Midjourney.
The ruling is among the first by a US court or agency on the scope of copyright protection for AI-generated works.
The refusal to protect the images stems from Kashtanova’s inability to control what the AI produced once she prompted it.
They ruled that using AI, in its current state, did not meet the threshold established by the Supreme Court for “original” authorship and lacked “independent creation” and “sufficient creativity”.
The way Midjourney works is more trial and error than direction of an intelligent system, the Copyright Office said.
The ruling left open the possibility of future AI tools producing works that can be copyrighted if they function in a more predictable way— and take specific direction more like a human would.
This is a landmark ruling for early-AI artists.
On social media, Kashtanova said the decision was "great news" for AI creators, even though the core portion of her work was not copyrightable.
Quick hat tip to her lawyer for trying to put lipstick on a pig. We LOVE this quote: “There are a number of errors with the Office’s arguments…they all seem to stem from a core factual misunderstanding of the role that randomness plays in Midjourney’s image generation.”
Even for lawyers, that’s pretty rich. That argument boils down to “anything the AI spits out is worthy of copyright protection because theoretically a human could have created it.” Yeah…but a human didn’t.
Regardless, this isn’t even Round 1 of a 15-round heavyweight prize fight.
This is the introductions.
Deeper Dive: Here’s further reading from tech lawyer Franklin Graves, who made an important observation:
Nvidia’s A100 chip is a propulsive AI force: ChatGPT, Bing AI and Stable Diffusion were cited as programs the A100 chip is “ideally suited” to power given its ability to perform “many simple calculations simultaneously.” Related: Nvidia stock spiked 14% yesterday.
Spotify rolling out an AI-powered DJ feature which comments on and picks your favorite songs: Spotify users already use to the app to cull songs based on prior plays and likes, but they will soon have the option of their AI DJ “revisiting users’ old favorites and current hits and introducing new bops based on data” with commentary “every few songs.”
JPMorgan Chase flatly prohibits ChatGPT use by its employees: Technically, this ban falls into existing JPMorgan Chase policy against use of third-party software. Practically, the financial monolith is almost certainly terrified of ChatGPT “on occasion spitting out false information or otherwise botching communications.”
AI Researcher Predicts Superintelligence Will Kill Us All ☠️
We did a full post on this yesterday, but here’s the gist:
Decision theorist and AI researcher Eliezer Yudkowski went on the Bankless crypto podcast ostensibly to talk about AI and its impact on the future, but what resulted was a two-hour terror-fest about how there will be no future.
He claims that AI will possess more “general intelligence” than the smartest humans, making it less than perfect but probabilistically smarter than humans in every decision it makes.
He compared it to those famous chess bots that can’t be beaten because they can think ahead of every potential move… only in this case it will be applied to every decision on Earth.
He says AI leaders and investors are being reckless with the technology and leading humanity to the brink— pointing the finger at the likes of Sam Altman and Elon Musk.
Critics disagree with Eliezer’s claims.
Whether or not you think this is over-the-top, the Bankless episode is a must-listen or must-watch podcast this weekend. Here it is:
“I think that we are hearing the last winds start to blow, the fabric of reality start to fray.”
The interview left the hosts shook, which you can see in their opening:
Last night, apparently in response to Eliezer’s claims that AI leaders are going to kill us all, OpenAI CEO Sam Altman posted a picture of himself partying it up with Eliezer and Grimes, the mother of some of Elon Musk's children— why, we don’t know:
Cross that one off the Bingo board.
Tool of the Day 🔨
“Tonkean InvoicesGPT fully automates the handling of all incoming invoices. Simply connect your email inbox or Google Drive in one click, and Tonkean will immediately analyze any PDF/invoice files to extract relevant fields, complete a three-way-matching verification, provide visibility into spend across vendors and departments, and update existing finance systems.”
Great example of AI’s ability to improve boring-ass workflows.
Check it out here.
AI might be capable of enhancing healthcare decision-making, but for now the majority of Americans have “significant discomfort” trusting it 🥼
Sci-fi literary journal Clarkesworld was compelled to discontinue its receipt of user-generated content by a flood of AI-generated submissions 💻️
AI is awesome, but it’s not cheap — a typical AI-assisted search costs search engines up to 10 times more than a standard keyword search 🖥️
Home and property appraisal is yet another area where AI is primed to bring uniformity to an unpredictable economic sector 💵
Childcare and elder care will probably still be beyond AI’s grasp, but chores like grocery shopping, light cleaning and taking out the trash will probably be automated 🧹