This isn't perfect. The cuts between stories are too quick; the writing is a little too formal; and the speech is just a bit flat.
But those are nits. Generally speaking, this is holy shit material. How long will it be before my blog is written by a computer?
See the highest quality AI footage in the world.
???? - Our generated anchors deliver stories that are informative, heartfelt and entertaining.
Watch the showcase episode of our upcoming news network now. pic.twitter.com/61TaG6Kix3
— Channel 1 (@channel1_ai) December 12, 2023
They keep to the same mannerisms as if in a random loop, which then causes their body actions to be out of sync (actions that seem at odds with what they're saying) at odd times with their mouths/voice. It's not just that their speech is a bit flat, but lacks emotion -- the pace is steadier than how humans actually talk, and it's delivered close to monotone.
There's also the single camera shot, which feels odd given that almost every news broadcast makes use of multiple angles and zooms.
There's something too solid seeming about the AI images, they obviously aren't going to go off-script.
Yes, I bet there are human editors. Also, humans must give quite a bit of input to the AI tool describing what they want as the output. So it's not just AI turned loose with little direction.
yeah, noticed the repeated hand movement immediately, BUT i bet without already knowing they were AI i would have been fooled. and it's going to get a lot better. as kevin said, the problems we're seeing are nits to be picked.
Are there any articles about how we transition to the new AI economy? Obviously a lot of people like me (newspaper copy editor) are going to lose our jobs. How are we going to cope? Are there plans in place for this?
I find Kevin's lack of concern about this to be rankling — and I say this as a longtime fan of his blog. He's retired already so he has no personal investment, but the AI economy is clearly going to be destabilizing. How are the rest of us going to get by?
I think if you go back through his posts over the past couple of years, Kevin has written quite a lot about how we're not concerned *enough* over how disruptive AI will be.
Was just about to make a similar comment. Kevin if anything has shown far more "concern" about the replacement of human worker than typical elite pundits, most of whom accept the neoliberal orthodoxy about ever-growing opportunities for human workers
How are we going to cope? Are there plans in place for this?
Narrator (probably powered by AI): You're not. There are no plans in place for this, and the Powers That Be just hope to distract you with nonstop bread (shit that doesn't matter)* and circuses (shit that is legitimately infuriating)** long enough until it's too late for you to do anything legal*** about it and you just die or join the ranks of everyone else begging for scraps from the owner class.
*Trans people using the "wrong" bathroom, Critical Race Theory taking over our schools, drag queens indoctrinating our kids, Christian bakers/coaches/website designers being persecuted, caravans of "illegals" approaching the border and then being bused all over the country, the national debt needing to be paid off right now, etc.
**Attempts to overthrow our government and the rule of law, voter disenfranchisement, Kafkaesque court rulings, gun violence everywhere, rights being repealed, corporations recording record profits as they jack up prices, climate change inaction, non-stop criming by the Republican (and occasional Democrat) elite, etc.
***The hilarious part is that the Powers That Be assume that all the guns lying around the country won't be turned on them when Shit Gets Real for a critical mass of Americans and Civil War II erupts.
It's called the Butlerian Jihad and Frank Herbert predicted it in Dune. It's kinda inevitable. The only thinking machines that survived were simple ones like environmental controls and telemetry. "Thou shalt not make a machine in the likeness of a human mind." Instead, mentat philosophy, drugs, implants and skills were invented so that people could serve as computers.
"It's called the Butlerian Jihad and Frank Herbert predicted it in Dune"
Herbert predicted nothing. He wanted to write a story about humans in an arbitrarily far future, not robots, and so came up with a narrative device that explained why there were no computers. He certainly wasn't writing Dune as an exercise in prognositcation about AI.
(he did write Destination: Void, though).
I’m not going to watch that. Anyway… AI generated porn! That’s where the action is. ????
Anyway Mr. Al manifestation of the former Kevin Drum, we already know you’re not real.
I expect AI generated porn to become a massive social problem. What’s to stop people from creating it “starring” themselves and some person they find attractive? Then sharing it with friends? Or using it to bully people? It all seems scary to me.
@iamr4man
In short order, insane and impossible porn will become commonplace, and it won't be "scary," or even very offensive. Bout time, too.
How long will it be before my blog is written by a computer?
I thought it already was, using ChatKD-5
Charlie always uses the latest edition.
Haha
I posted this as my very tiny travel place and received some push back, here is my push back to that:
I though this Scary Good (taken from Kevin Drum):
...oh certainly, you are correct re hands and hand gestures are a little off...but I've watched this evolve from stick figures to faces where teeth were always somehow fundamentally incorrect as well as you noted, the complex problems with creating hands.
But they are almost really almost all the way there...they are not on the last mile in creating functioning human beings, but the last couple of feet.
A lot of jobs are going to be replaced....I have also favored immigration, in Britain as well as the US to avoid food inflation...
but see this BBC, where I saw it first, but here is a shorter version on CBS...damn these robots are getting good! I can tell that my tomatoes now come from such a robot green house...they are too perfect!
https://www.youtube.com/watch?v=1jxzkIjOUSw
Brave New World indeed!
Traveller
They showed a guy speaking French, but the French was translated into English. It's a miracle! Before 2023, there was never a way to translate French into English, so we could never watch French people say anything unless we were doing it just to laugh at the silly sounds they make.
We are so f*cked.
Why would anyone want this?
It's not so much a matter of why would anyone actively seek out AI-driven TV news. But rather, do people care one way or another if it's AI?
I suspect a lot of Americans are indifferent as long as it's accurate.
But more critically, that number is likely to grew once it becomes truly seamless. It's not far from that now, by the looks of this video.
I suspect a lot of Americans are indifferent as long as... it matches their preconceptions.
Was just about to make a similar comment. Kevin if anything has shown far more "concern" about the replacement of human worker than typical elite pundits, most of whom accept the neoliberal orthodoxy about ever-growing opportunities for human workers
The advances in AI are impressive....and it has the same problem as all new aggregator sites. It does not investigate, and as any good business model would, it probably does not pay, or pay much, to those who do the investigation. Leading to the dearth and death of local news everywhere.
". . . News just the way you want it -- personalized, localized . . ."
Just what we don't need more of: personalized news.
I haven't yet seen a solution for the "creating false information" problem you highlighted last week. I'm sure they're working on it (ChatGPT was better than Bard.)
I only see three possibilities: 1) some kind of automated filtering/fact-checking editing process, 2) human fact checking, or 3) full semantic cognition. #3 seems a pretty big leap (basically the sci-fi version of AI) from our already impressive capabilities, so I think that's a ways off, but I don't really know. #1 and #2 are probably the stopgaps that will be used.
Does anyone doubt Pinocchio is a real boy?
They've been telling me Santa isn't real for years, yet everywhere I look there's his picture!
Every year I ask people who believed in Santa whether finding he wasn’t real made them question their belief in God. No one I’ve asked said it did. I just don’t understand that.
But it really is his picture, isn't it?
Of course! Now go to bed and in the morning we’ll see where the elf on the shelf is.
Apparently, Michael Cohen's lawyer used some Chatbot to generated a brief, and it hallucinated three non-existant "precedent" cases ... which the judge cited him on.
Not the kind of entity that you want "reporting" news...
Ever since Dan Rather was fired there's been a total separation of the news person from responsibility for the news they present. This is the final nail in that coffin.
I'm not saying we can control the future. That way lies madness. Just look at Speaker Mike and the Donald and Ronald twins.
Nor am I saying there's nothing we can do. Instead, I am saying something that seems to be incredibly radical, which is that time, energy, and attention precedes understanding.
I'll refer to artificial intelligence as +AI, and artificial ignorance as -AI.
-AI is acting like people that think they can control others, or people that give up and thereby enable the controllers. The controllers and enablers are detached from reality in that the only human being you can control is you.
To be more specific, we can constrain others if we are willing to be constrained. The only question is, what is the constraint?
I'll use scientific language based on the assumption most people reading this identify with science. The scientific principle involves careful observation, applying rigorous skepticism about what is observed, given that cognitive assumptions can distort how one interprets the observation. I got that from Wikipedia's Scientific Method article, but I'm using it because I think it hits the bullseye.
Our response to +AI is on the other side of understanding. Understanding takes time, energy, attention, and a baseline level of intelligence. If you have a valid kindergarten graduation certificate, then you meet that baseline.
Bottom line, don't give up, and don't try to control. Instead, constrain, and start by constraining yourself.
For example, say two people have conflicting conclusions about how to respond to +AI. Now imagine that they are both willing to behave within the "scientific principle" constraint.
Instead of insisting "I'm right and you're wrong," they're having an adult conversation. They don't naively accept, and they don't cynically dismiss the other person's conclusion. They do defend their assumptions, but they allow the other person to challenge their assumptions up to and including the willingness to recognize and acknowledge "my bad."
If you identify with a religion, or capitalism, or freedom, then the language/medium is different, but the message is identical.
If the issue is +AI, or abortion, or LGBTQIA2S, or the economy, or crime, or boarder security, or the Ukrainian crisis, or Middle East crisis, or political discord, or who to vote for in 2024, then my words do not have to change.
P.S.
If there's a person in a position of power in the MAGA Republican party with a valid kindergarten graduation certificate, they're hiding.
So, has anyone created an AI Jesus yet? WWJD? Just ask him! That might be interesting. How about an AI God?
I suspect an attempt to create an AI Trump would just cause the computer to melt down.
I think the appeal of Trump has a lot to do with the idea that what he says is not what a politician would be programmed to say. We live in an era of over-prepared politicians who haven't uttered a spontaneous thought since they were in elementary school. Long ago, they learned the algorithm of what politicians are supposed to do. Some do it ineptly, like DeSantis. Some more artfully, like Nikki Haley. They take their cues from the class of consultants and journalists who think they know how to appeal to an audience but don't.
Trump is not what any AI programmer would come up with if they were building an AI politician. He's coarse, he's uninformed, he's insulting. He lies all the time. Yet according to an in-law of mine, he "tells it like it is." Why? Because he doesn't sound like every other politician we've listened to for the past 30 or 40 years.
Biden doesn't sound like other pols either. He can be inept on the stump. It's not that he has succeeded despite his limitations, but because of them. He comes across as real. When he's sincere and heartfelt, you feel it. He's AI-proof.
Maybe that's why older pols rule today. (Sanders another one.) They have some memory of a time that seemed more real. Today's generations grew up on media, primarily television, the most artificial medium in human history.
That's why the AI news video here seems so uninvolving to me. It does a whiz-bang job of simulating a typical "real" news broadcast, but TV itself is such a faux medium to begin with, my reaction is, So what if you can recreate what's already fake?
Such diversity in the newscasters! Not a single one appeared over the age of 40.
This is a sizzle reel designed to sell a product. Even recent releases from companies like Meta have repeatedly turned out to be fake implementations for real systems that don't actually do what was claimed in their roll-out event. How many human beings worked for how many hours tweaking or rerunning or otherwise adjusting this video? Short of them showing somebody typing in a story live and then having the AI render itself in real time, it's unclear how much this actually represents what their AI can do.
It's also unclear what kind of support structure will be needed: if this AI simply scrapes the Internet for news, or accepts whatever text it is fed, then it doesn't replace reporting or fact-checking or any of the parts of a news-team except for the anchors.
My big concern is that AI journalists aren't trained in the same way. Can we have a conversation with this AI about journalistic ethics, for example? There's certainly unethical human journalists, but there's a big difference between buying up hundreds of media outlets and expecting every single employee to broadcast the news you dictate to them without word getting out, and buying up an AI news broadcaster and watching a bunch of techs maintain and operate the thing without any idea what journalistic ethics are or how they're being asked to maintain an AI that will casually violate them.
Bezos can't tell the Washington Post what news to publish: he has influence, but there's too many people involved in the process to interfere beyond a limited point. But the AI Post will do exactly as instructed without complaint.
Genuine AI could make a big difference: real-time language translation or closed captioning on your phone, companions and assistants for the elderly, AI-operated vehicles, AI-assisted design. But other forms of AI promise to give more power to the wealthy few without any accountability. I suspect that, at least in all our lifetimes, that promise will be more of a con, but I'm concerned about the ethics of those designing and selling these AI, as well as their lack of diversity and their disregard for anyone unlike themselves. Just watching my elderly relatives increasingly struggle with all the "convenient tech" available to them makes clear that the needs of people who aren't middle-age techies aren't being considered much in the development process.
The apocryphal story about the Lumiere brothers' first presentation of short films in 1895 was that members of the audience, seeing a train pulling into a station, jumped out of their seats. They didn't know the train was a moving image on a screen. They thought the train was real and coming down the tracks right at them.
Leaps in technology often "fool" people. The audience confuses what's real and what's not real. For a time, perhaps, then people adapt. They catch on to the manipulation. They see the man behind the curtain. Or the puppet strings, even if they're well hidden.
People don't reject the technology of manipulation. They want more of it, but they want it to be done well. They want it to seem real. They want it to surprise them. All of that is fine. If it's Walt Disney making kids laugh, or Stephen Spielberg recreating a story of the Holocaust, or George Lucas inventing a space age adventure, what's the harm?
Maybe the harm is when we don't know whether to suspend our disbelief or to take in what we're told with a critical eye. The motivation of the audience has a lot to do with it. Right now, we're hearing two sides about a conflict in the Middle East, both offering competing narratives with a mixture of truths and half-truths, and a disregard for the other side. For many, what matters is not what really happened so much as what feeds their desire for a certain "truth." That's how it is with much political debate and wherever you find conflict between people.
The more motivated the audience, the easier it is to manipulate them. That's nothing new. It was 1897 when William Randolph Hearst sent a cable to his man in Cuba: "You furnish the pictures and I'll furnish the war." His man was not an AI programmer or a filmmaker or a photographer, but an artist. Hearst got his war anyway.
Trump is a master manipulator, but extremely low tech. All he does is stand at a mike and talk. Most people here see through the lies. We're immune to his manipulation. But his audience doesn't care. They want to believe like kids watching Mickey Mouse or Luke Skywalker.
We're all subject to manipulation in one way or another. But some of us are a lot less oblivious to the truth. I'm not sure how technology will help us lean toward the side of what's real and away from what's not, but I hope that artists and journalists and historians and teachers are working on that. Soon.
"We're all subject to manipulation ... "
Please speak for yourself. I explained how I avoid being subject to manipulation in a previous comment on this same post. If you mean "We (who ignore the scientific principle as much too inconvenient) are all subject to manipulation ... " then I agree.
"I hope that artists and journalists and historians and teachers are working on that."
I think what you're saying is, "Something must change, but there's no way it's going to be me." I'm mostly skeptical but still a little openminded to the idea I got that wrong.
Somewhere in a digital padded cell, Max Headroom is kicking the walls and screaming Now! It's my time! Now!
As another AI problem, China won't stand by and have AI simply gather the facts and strive to tell the truth. They'll only allow the creation of AI that gives out information that conforms to the current communist partly line. If we assume that AI will somehow only search for the truth, we are kidding ourselves. The human AI algorithm developers can create AI with all sorts of biases. So we have AI ideological wars ahead.
And I know we all can't wait for the MAGA version of AI!
Well, the current AI doesn’t know what is truth. It doesn’t “know” anything. It is PCA (Principle Component Analysis) on steroids.
It will be interesting to see what happens when today’s AI generated content becomes the dominant content on the web. Then we’ll have the algorithms eating themselves. Fun times.
I’ve seen articles about how scientists are generating content that when consumed by AI screws up their databases. Interesting times indeed.
how many entry level jobs of any given "white collar" profession will remain? So, just a further tightening of the funnel, so that a newsroom, law firm, etc really doesn't have any use for early employees that wash out.
We are so unprepared for this.
If this news program was completely done by AI, one of the more impressive aspects are the humor bits during the movie segment starting c. 17:00. There are a few decent chuckles in there.
This is terrible bullshit and Kevin Drum should be embarrassed at being impressed (?) by this. It's anti-human gibberish and the only difference is that the computers are much cheaper at making the same anti-human gibberish that humans have been making. It's a visual pablum-and-platitude generator.
"It's a visual pablum-and-platitude generator."
There's absolutely a market for that, and if AI is just as good at creating it as humans are, and at much lower cost, then our current (human) platitude generators will soon be SOL. Along with lawyers, software engineers, and gawd only knows how many other professions who will also be replaced by AI.
That's why people are impressed.
How can a story generated by a machine be "heartfelt"?
How about tomorrow? See what it can do. Give it topics and see what comes out, without edits.
I came back to this post to watch more of the sample. After about 7 minutes, somehow I started to feel kind of sick -- not because it was too real but because there was something very unsettling about the mannerisms of these fake people. Not just "flat" as Kevin says, but somehow the lack of human feeling just breaks through. The pacing, the evenness of the pauses, the way the vocal pitch changes seem to have no relation to the story content (Gaza bombing sounds no different from the Bears-Panthers game), things like that.
It's not exactly boring, but somehow off-putting in a way that I have also gotten from watching actual human newsreaders that obviously don't care at all about what they're reporting. Only here it's only more so.
Maybe I wouldn't feel this way if I didn't KNOW they were fake, but I don't think so. Somehow the effort to make them so realistic makes the feeling of emptiness even worse than if they looked and sounded like robots.