[News] The AI Thread!

News updates on the development and ramifications of AI. Obvious header joke is obvious.

As much as I am not inclined to think outlawing vice works in the end (tons of nuance here don't get me wrong just laying out my broad position) it's wild to me that there's anyone that's not fully aware of the fact this isn't a dilemma , it's just bad? There's no debate here it's always been about profits with no regard towards the welfare of people beyond what they have been forced to do. To say otherwise strikes me as willfully naive in that "If I state the facts I risk a lawsuit" kinda way

So Elon sued OpenAI.

(The Verge link, they are rather skeptical of Musk's suit, at best.)

Elon Musk sued OpenAI today, alleging a wide range of incendiary things, including that GPT-4 is actually an artificial general intelligence. It’s a fun complaint to read; it fundamentally accuses OpenAI and its CEO, Sam Altman, of pretending to run a nonprofit designed to benefit humanity while actually running a regular ol’ tech company and trying to make a lot of money. That’s a pretty good criticism of the entire OpenAI situation, actually! Someone with some intellectual honesty and a competent lawyer should run at that sometime.

Sadly, Musk is not that person, and his lawyers have figured out that letting the world’s richest man rack up billable hours filing nonsensical lawsuits is more lucrative than fitting the “facts” to the “law,” or whatever it is regular lawyers do.

Let’s just take the very first cause of action of the lawsuit, for example. It is a claim for breach of contract — a very, very simple claim that almost any first-year law student can evaluate, because step one is asking if there is a contract, and step two is figuring out what the contract says. To have a valid contract, you need an offer, acceptance, and an exchange of value — what lawyers are trained to call “consideration,” in an enduring effort to make simple concepts sound confusing and increase fees.

Most importantly, contracts need to be written down — proving that an unwritten contract exists, what its terms are, and if they are enforceable is extraordinarily difficult, and courts do not like doing it, especially for ultra-sophisticated parties with a long history of dealing.

My friends, Musk is straightforwardly alleging that OpenAI breached a contract that does not exist. It is simply not a thing! The complaint makes reference to a “Founding Agreement,” but no such Founding Agreement is attached as an exhibit, and the breach of contract claim admits that the “Founding Agreement” is basically a vibe everyone caught in some emails.

Anyway, OpenAI fired back yesterday:

OpenAI says Elon Musk wanted ‘absolute control’ of the company

OpenAI has responded to Elon Musk’s lawsuit by saying that he at one point wanted “absolute control” of the company by merging it with Tesla.

In a blog post published on Tuesday, OpenAI said it will move to dismiss “all of Elon’s claims” and offered its own counter-narrative to his account of the company abandoning its original mission as a nonprofit.

“As we discussed a for-profit structure in order to further the mission, Elon wanted us to merge with Tesla or he wanted full control,” including “majority equity, initial board control, and to be CEO,” according to the post, which is authored by OpenAI co-founders Greg Brockman, Ilya Sutskever, John Schulman, Sam Altman, and Wojciech Zaremba. “We couldn’t agree to terms on a for-profit with Elon because we felt it was against the mission for any individual to have absolute control over OpenAI.”

Like, I may not like OpenAI, but that sounds completely plausible for King Dipsh*t.

Microsoft asks to dismiss New York Times’s ‘doomsday’ copyright lawsuit

Microsoft has responded to a copyright infringement lawsuit brought by the New York Times over alleged use of content to train generative artificial intelligence, calling the claim a false narrative of “doomsday futurology”. The tech giant said the lawsuit was near-sighted and akin to Hollywood’s losing backlash against the VCR.

In a motion to dismiss part of the lawsuit filed on Monday, Microsoft, which was sued in December alongside ChatGPT-maker OpenAI, scoffed at the newspaper’s claim that Times content receives “particular emphasis” and that tech companies “seek to free-ride on the Times’s massive investment in its journalism”.

In the lawsuit – which could have major implications for the future of generative artificial intelligence and for news-content production – the Times alleged that Microsoft, which is OpenAI’s biggest investor, had unlawfully used the paper’s “copyrighted news articles, in-depth investigations, opinion pieces, reviews, how-to guides, and more” to create artificial intelligence products that “threatens The Times’s ability to provide that service”.

But in its response, Microsoft said the lawsuit was akin to Hollywood’s resistance to the VCR that consumers used to record TV shows and which the entertainment business in the late 1970s feared would destroy its economic model.

“‘The VCR is to the American film producer and the American public as the Boston strangler is to the woman home alone,’” Microsoft said in its response, quoting from congressional testimony delivered by Jack Valenti, then head of the motion picture association of America, in 1982. In this case, Microsoft said, the Times was attempting to use “its might and its megaphone to challenge the latest profound technological advance: the Large Language Model.” Microsoft’s lawyers also argued that “content used to train LLMs does not supplant the market for the works, it teaches the models language”.

OpenAI has already asked a judge to dismiss parts of the lawsuit against it, alleging that the publisher “paid someone to hack OpenAI’s products” to create examples of copyright infringement using its ChatGPT.

“ChatGPT is not in any way a substitute for a subscription to The New York Times,” attorneys for OpenAI wrote. “In the real world, people do not use ChatGPT or any other OpenAI product for that purpose. Nor could they. In the ordinary course, one cannot use ChatGPT to serve up Times articles at will.”

But the battle comes against a series of lawsuits brought by authors and artists over various aspects of copyright, including ownership of creative work created using the technology, and complaints that AI technology can create some wildly misleading information, what the industry cutely calls “hallucinations”.

Last month, Google was forced to apologize when its Gemini chatbot was used to create images of Black soldiers in second world war-style German military uniforms and Vikings in traditional Native American dress. Google temporarily suspended the technology’s ability to make images of people and vowed to fix what it described as “inaccuracies in some historical” depictions.

The twin concerns – that AI technology can violate copyrighted material and create information or images that are wildly improbable – comes as OpenAI recently acknowledged that it was “impossible” to train AI models without copyrighted works “because copyright today covers virtually every sort of human expression”. OpenAI has refused to disclose the contents of its training databases, including for its newest tool, a video generator called Sora.

In a letter to the UK’s House of Lords, the company also said that “limiting” the training data to content in the public domain “would not provide AI systems that meet the needs of today’s citizens”.

The OpenAI CEO, Sam Altman, said in January that he was “surprised” by the Times’s lawsuit because the system did not need the Times data to train itself. “I think this is something that people don’t understand. Any one particular training source, it doesn’t move the needle for us that much.” Altman claimed the Times’s articles represented a minute part of the corpus of text used to create ChatGPT.

Filing frivolous lawsuits is Musk's new thing.

AI Marilyn Monroe adds to the list of dead celebrities digitally resurrected without consent

Announced on International Women's Day, no less.

ChatGPT uses 17,000 times the amount of electricity than the average US household does daily: report

This is an educated guess, but so far OpenAI hasn't denied it. Perhaps the future envisioned in The Matrix isn't so far off.

Midjourney is mad that rival ai company Stabiity AI was apparently scraping their images after "botnet-like activity" caused a 24-hour outage, and has banned all Stabiity AI employees from using Midjourney.

IMAGE(https://www.themarysue.com/wp-content/uploads/2021/05/spider-man-point-meme.jpg)

https://www.schneier.com/blog/archiv...

Researchers ran a global prompt hacking competition, and have documented the results in a paper that both gives a lot of good examples and tries to organize a taxonomy of effective prompt injection strategies.

Bit of a shower thought: AI will probably not take very long to 'out produce' human art (and text etc). For example a streamer I watch has a fan that set up a bot for subscribers that inserts the username into a detailed prompt that fits the community theme and gives the subscriber a sort of avatar image. There are thousands of these now. I'm picturing a few decades from now there are billions of pieces of AI art training more on each other than on human input at that point. Where does that end up? Given a thousand years will all AI art output be a grey square or what?

polypusher wrote:

Bit of a shower thought: AI will probably not take very long to 'out produce' human art (and text etc). For example a streamer I watch has a fan that set up a bot for subscribers that inserts the username into a detailed prompt that fits the community theme and gives the subscriber a sort of avatar image. There are thousands of these now. I'm picturing a few decades from now there are billions of pieces of AI art training more on each other than on human input at that point. Where does that end up? Given a thousand years will all AI art output be a grey square or what?

Instead of having to click the squares with the motorcycle to prove you’re human you’ll have have to click the squares with art.

Beauty is in the eye of the beholder. Who will want a big grey square? The bigger question is, who will be the beholder.

polypusher wrote:

Bit of a shower thought: AI will probably not take very long to 'out produce' human art (and text etc). For example a streamer I watch has a fan that set up a bot for subscribers that inserts the username into a detailed prompt that fits the community theme and gives the subscriber a sort of avatar image. There are thousands of these now. I'm picturing a few decades from now there are billions of pieces of AI art training more on each other than on human input at that point. Where does that end up? Given a thousand years will all AI art output be a grey square or what?

Model collapse.

Seeing an AI guy on Twitter react to Google's new Vlogger tool by actually saying "Now anybody can become a YouTuber" is just too perfect.

Prederick wrote:

Seeing an AI guy on Twitter react to Google's new Vlogger tool by actually saying "Now anybody can become a YouTuber" is just too perfect.

The thread of responses that follows is the first half of a Black Mirror episode.

Nvidia: what’s so good about the tech firm’s new AI superchip?

What is Nvidia doing?

The main announcement of the company’s annual develop conference on Monday was the “Blackwell” series of AI chips, used to power the fantastically expensive datacentres that train frontier AI models such as the latest generations of GPT, Claude and Gemini.

Alphabet's Google Releases Gemini AI Model
The Gemini logo on a smartphone arranged in New York, US, on Saturday, Dec. 9, 2023. Alphabet's Google said Gemini is its largest, most capable and flexible AI model to date, replacing PaLM 2, released in May. Photographer: Gabby Jones/Bloomberg via Getty Images

One, the Blackwell B200, is a fairly straightforward upgrade over the company’s pre-existing H100 AI chip. Training a massive AI model, the size of GPT-4, would currently take about 8,000 H100 chips, and 15 megawatts of power, Nvidia said – enough to power about 30,000 typical British homes.

With the company’s new chips, the same training run would take just 2,000 B200s, and 4MW of power. That could lead to a reduction in electricity use by the AI industry, or it could lead to the same electricity being used to power much larger AI models in the near future.

What makes a chip ‘super’?

Alongside the B200, the company announced a second part of the Blackwell line – the GB200 “superchip”. It squeezes two B200 chips on a single board alongside the company’s Grace CPU, to build a system which, Nvidia says, offers “30x the performance” for the server farms that run, rather than train, chatbots such as Claude or ChatGPT. That system also promises to reduce energy consumption by up to 25 times, the company said.

Putting everything on the same board improves the efficiency by reducing the amount of time the chips spend communicating with each other, allowing them to devote more of their processing time to crunching the numbers that make chatbots sing – or, talk, at least.

‘Fraud is fraud’: Georgia aims to ban AI deepfakes in political campaigns

When wrangling legislation, sometimes it’s best to sound out a problem in front of you.

In Georgia, it sounds like the state senator Colton Moore. But it only sounds like Colton Moore.

Todd Jones, a Republican state representative who chairs the Georgia house committee on technology and infrastructure innovation, has proposed legislation outlawing the use of artificial intelligence deepfakes in political communication. To illustrate the point, Jones presented a deepfake video to the judiciary committee using an AI image and audio of Moore and Mallory Staples, a former Republican congressional candidate who now runs a far-right activist organization, the Georgia Freedom caucus.

The video uses an AI tool to impersonate the voices of Moore and Mallory falsely endorsing passage of the bill. The video contains a continuous disclaimer at the bottom citing the text of the bill.

Moore and Mallory oppose the legislation.

The AI impersonation of Moore says: “I would ask the committee: how is using my biometric data, like my voice and likeness, to create media supporting a policy that I clearly don’t agree with the first amendment right of another person?”

The video continues: “The overwhelming number of Georgians believe the use of my personal characteristics against my will is fraud, but our laws don’t currently reflect that. If AI can be used to make Colton Moore speak in favor of a popular piece of legislation, it can be used to make any one of you say things you’ve never said.”

Brad Thomas, the Republican co-sponsor of the bill and co-author of the video, said he and his colleagues used commonly available tools to create the video.

“The particular one we used is, like, $50. With a $1,000 version, your own mother wouldn’t be able to tell the difference,” he said.

The pace of advancement of visual AI generative tools is years ahead of the legislation needed to prevent abuses, Thomas said: “Cinematography-style video. Those individuals look absolutely real, and they’re AI-generated.”

The bill passed out of committee on an 8-1 vote.

Moore is not popular in Georgia’s legislative circles. His peers in the state senate threw him out of the Republican caucus in September, accusing him of making false statements about other conservatives while he was advocating fruitlessly for a special session to remove the Fulton county prosecutor Fani Willis from office.

Last week, Moore was permanently barred from the Georgia house chamber after rhetorically attacking the late speaker at a memorial service being held on the house floor.

Through the Georgia senate press office, Moore declined to comment.

In social media posts, Moore has voiced opposition to this bill, which he said is an attack on “memes” used in political discourse, and that satire is protected speech.

Staples, in newsletters to her supporters, cited the federal conviction of Douglass Mackey last year as an example of potential harms. Mackey, also known as the alt-right influencer “Rickey Vaughn”, sent mass text messages in November 2016 encouraging Black recipients to “vote by text” instead of casting a real vote, with the texts claiming they had been paid for by the Clinton campaign.

Federal judges rejected Mackey’s first amendment arguments on the ground that the communications amounted to acts of fraud which were not constitutionally protected. Mackey was sentenced in October to serve seven months.

House bill 986 creates the crimes of fraudulent election interference and soliciting fraudulent election interference, with penalties of two to five years in prison and fines up to $50,000.

If within 90 days of an election, a person publishes, broadcasts, streams or uploads materially deceptive media – defined as appearing to depict a real individual’s speech or conduct that did not occur in reality and would appear to a reasonable person to be authentic – they would be guilty of a felony, as long as the media in question significantly influences the chances for a candidate or referendum to win, or confuses the administration of that election. Thus, it would also criminalize using deepfakes used to cast doubt on the results of an election.

Deepfakes entered the 2024 election at its start, with an AI-generated audio call featuring Joe Biden telling New Hampshire voters not to vote. After the call, the Federal Communications Commission announced a ban on robocalls that use AI audio. But the Federal Elections Commission has yet to put rules in place for political ads that use AI, something watchdog groups have been calling for for months. Regulations are lagging behind the reality of AI’s capabilities to mislead voters.

In the absence of federal elections rules for AI content, states have stepped in, filing and, in several instances, passing bills that typically require labels on political ads that use AI in some way. Without these labels, AI-generated content in political ads is considered illegal in most of the bills filed in states.

Experts say AI audio, in particular, has the ability to trick voters because a listener loses context clues that might tip them off that a video is fake. Audio deepfakes of prominent figures, such as Trump and Biden, are easy and cheap to make using readily available apps. For less well-known people who often speak publicly and have a large volume of examples of their voices, like speeches or media appearances, people can upload these examples to train a deepfake clone of the person’s voice.

Enforcement of the Georgia law might be challenging. Lawmakers struggled to find ways to rein in anonymous flyers and robocalls spreading misinformation and fraud ahead of elections long before the emergence of AI.

“I think that’s why we gave concurrent jurisdiction to the attorney general’s office,” Thomas said. “One of the other things we’ve done is allow the [Georgia bureau of investigation] to investigate election issues. Between the horsepower of those two organizations, we have the highest likelihood of figuring out who did it.”

Lawmakers are only just starting to get at the implications of AI. Thomas expects more legislation to emerge over the next few sessions.

“Fraud is fraud, and that’s what this bill is coming down to,” Thomas said. “That’s not a first amendment right for anyone.”

Rachel Leingang contributed reporting

Old People Are Praying To An AI Shrimp Jesus On Facebook

IMAGE(https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,format=auto,onerror=redirect,quality=80/uploads/asset/file/6973a7cb-05f2-4b68-88dd-c483660c68af/Screenshot_2024-03-20_at_2.35.38_PM.png?t=1710959753)

I was tipped off about this earlier this month, but I didn’t cover it because I felt like I had just covered this recently. But, luckily, the good folks over at 404 Media picked the story up.

What’s happening here is actually two different content hacks occurring at the same time.

As I wrote in October, the “Amen” comment is a new growth hack on Facebook. You get people to say “Amen,” underneath a post, the post gets recommended to others, who say “Amen,” as well, etc. This is especially powerful on the platform right now because as news has been deprioritized on the platform, a lot more religious content has bubbled to the surface. This is why one the largest publishers on the site for the last six months has been a website literally just called catholicfundamentalism.com. it’s getting more engagement on Facebook, according to Newswhip, than, literally, every other publisher on Earth.

The second hack happening here is just the use of AI images. I had forgotten this, but back in the same issue I wrote about the “Amen” hack, I also wrote about what was likely patient zero for the AI image growth hack. It’s a page called “Uncle Mike's Photography” and it went super viral last year after publishing a bunch of AI images of men fighting gators in swamps. There’s also just the simple fact that Meta isn’t really trying to clamp down on this sort of thing. Meta’s chief apologizer, Nick Clegg, didn’t announce that the platform would start labeling AI-generated images until last month. And even then, Clegg said they’ll only label AI images when they can be auto-detected.

Which means, the bad food magicians that torment your grandparents will most likely soon be synthetic. Another job, lost to the machines.

these show up in my feed somewhat regularly due to older relatives and in my experience the comments are usually about 1/4 people complementing the post and 3/4 people calling it out for being AI.
It’s still engagement though.

Yeah, I feel like this about the absolute garbage engagement bait posts from entirely mainstream, real places i'll see on IG. They don't care if everyone in the comments is going "this is awful content slop, stop posting it you're total sh*t" because all the algo hears is "Wow! Look at all of that engagement!"

A ChatGPT for Music Is Here. Inside Suno, the Startup Changing Everything

"I'm just a soul trapped in this circuitry.”

The voice singing those lyrics is raw and plaintive, dipping into blue notes. A lone acoustic guitar chugs behind it, punctuating the vocal phrases with tasteful runs. But there’s no human behind the voice, no hands on that guitar. There is, in fact, no guitar. In the space of 15 seconds, this credible, even moving, blues song was generated by the latest AI model from a startup named Suno. All it took to summon it from the void was a simple text prompt: “solo acoustic Mississippi Delta blues about a sad AI.” To be maximally precise, the song is the work of two AI models in collaboration: Suno’s model creates all the music itself, while calling on OpenAI’s ChatGPT to generate the lyrics and even a title: “Soul of the Machine.”

Online, Suno’s creations are starting to generate reactions like “How the f*ck is this real?” As this particular track plays over a Sonos speaker in a conference room in Suno’s temporary headquarters, steps away from the Harvard campus in Cambridge, Massachusetts, even some of the people behind the technology are ever-so-slightly unnerved. There’s some nervous laughter, alongside murmurs of “Holy sh*t” and “Oh, boy.” It’s mid-February, and we’re playing with their new model, V3, which is still a couple of weeks from public release. In this case, it took only three tries to get that startling result. The first two were decent, but a simple tweak to my prompt — co-founder Keenan Freyberg suggested adding the word “Mississippi” — resulted in something far more uncanny.

Over the past year alone, generative AI has made major strides in producing credible text, images (via services like Midjourney), and even video, particularly with OpenAI’s new Sora tool. But audio, and music in particular, has lagged. Suno appears to be cracking the code to AI music, and its founders’ ambitions are nearly limitless — they imagine a world of wildly democratized music making. The most vocal of the co-founders, Mikey Shulman, a boyishly charming, backpack-toting 37-year-old with a Harvard Ph.D. in physics, envisions a billion people worldwide paying 10 bucks a month to create songs with Suno. The fact that music listeners so vastly outnumber music-makers at the moment is “so lopsided,” he argues, seeing Suno as poised to fix that perceived imbalance.

Most AI-generated art so far is, at best, kitsch, à la the hyperrealistic sci-fi junk, heavy on form-fitting spacesuits, that so many Midjourney users seem intent on generating. But “Soul of the Machine” feels like something different — the most powerful and unsettling AI creation I’ve encountered in any medium. Its very existence feels like a fissure in reality, at once awe-inspiring and vaguely unholy, and I keep thinking of the Arthur C. Clarke quote that seems made for the generative-AI era: “Any sufficiently advanced technology is indistinguishable from magic.” A few weeks after returning from Cambridge, I send the song off to Living Colour guitarist Vernon Reid, who’s been outspoken about the perils and possibilities of AI music. He notes his “wonder, shock, horror” at the song’s “disturbing verisimilitude.” “The long-running dystopian ideal of separating difficult, messy, undesirable, and despised humanity from its creative output is at hand,” he writes, pointing out the problematic nature of an AI singing the blues, “an African American idiom, deeply tied to historical human trauma, and enslavement.”

Suno is barely two years old. Co-founders Shulman, Freyberg, Georg Kucsko, and Martin Camacho, all machine-learning experts, worked together until 2022 at another Cambridge company, Kensho Technologies, which focused on finding AI solutions to complex business problems. Shulman and Camacho are both musicians who used to jam together in their Kensho days. At Kensho, the foursome worked on a transcription technology for capturing public companies’ earnings calls, a tricky task given the combination of poor audio quality, abundant jargon, and various accents.

Along the way, Shulman and his colleagues fell in love with the unexplored possibilities of AI audio. In AI research, he says, “audio in general is so far behind images and text. There’s so much that we learn from the text community and how these models work and how they scale.”

The same interests could have led Suno’s founders to a very different place. Though they always intended to end up with a music product, their earliest brainstorming included an idea for a hearing aid and even the possibility of finding malfunctioning machinery through audio analysis. Instead, their first release was a text-to-speech program called Bark. When they surveyed early Bark users, it became clear that what they really wanted was a music generator. “So we started to run some initial experiments, and they seemed promising,” Shulman says.

Suno uses the same general approach as large language models like ChatGPT, which break down human language into discrete segments known as tokens, absorb its millions of usages, styles, and structures, and then reconstruct it on demand. But audio, particularly music, is almost unfathomably more complex, which is why, just last year, AI-music experts told Rolling Stone that a service as capable as Suno’s might take years to arrive. “Audio is not a discrete thing like words,” Shulman says. “It’s a wave. It’s a continuous signal.” High-quality audio’s sampling rate is generally 44khz or 48hz, which means “48,000 tokens a second,” he adds. “That’s a big problem, right? And so you need to figure out how to kind of smoosh that down to something more reasonable.” How, though? “A lot of work, a lot of heuristics, a lot of other kinds of tricks and models and stuff like that. I don’t think we’re anywhere close to done.” Eventually, Suno wants to find alternatives to the text-to-music interface, adding more advanced and intuitive inputs — generating songs based on users’ own singing is one idea.

OpenAI faces multiple lawsuits over ChatGPT’s use of books, news articles, and other copyrighted material in its vast corpus of training data. Suno’s founders decline to reveal details of just what data they’re shoveling into their own model, other than the fact that its ability to generate convincing human vocals comes in part because it’s learning from recordings of speech, in addition to music. “Naked speech will help you learn the characteristics of human voice that are difficult,” Shulman says.

One of Suno’s earliest investors is Antonio Rodriguez, a partner at the venture-capital firm Matrix. Rodriguez had only funded one previous music venture, the music-categorization firm EchoNest, which was purchased by Spotify to fuel its algorithm. With Suno, Rodriguez got involved before it was even clear what the product would be. “I backed the team,” says Rodriguez, who exudes the confidence of a man who’s made more than his share of successful bets. “I’d known the team, and I’d especially known Mikey, and so I would have backed him to do almost anything that was legal. He’s that creative.”

Rodriguez is investing in Suno with the full knowledge that music labels and publishers could sue, which he sees as “the risk we had to underwrite when we invested in the company, because we’re the fat wallet that will get sued right behind these guys.… Honestly, if we had deals with labels when this company got started, I probably wouldn’t have invested in it. I think that they needed to make this product without the constraints.” (A spokesperson for Universal Music Group, which has taken an aggressive stance on AI, didn’t return a request for comment.)

Suno says it’s in communication with the major labels, and professes respect for artists and intellectual property — its tool won’t allow you to request any specific artists’ styles in your prompts, and doesn’t use real artists’ voices. Many Suno employees are musicians; there’s a piano and guitars on hand in the office, and framed images of classical composers on the walls. The founders evince none of the open hostility to the music business that characterized, say, Napster before the lawsuits that destroyed it. “It doesn’t mean we’re not going to get sued, by the way,” Rodriguez adds. “It just means that we’re not going to have, like, a f*ck-the-police kind of attitude.”

Rodriguez sees Suno as a radically capable and easy-to-use musical instrument, and believes it could bring music making to everyone much the way camera phones and Instagram democratized photography. The idea, he says, is to once again “move the bar on the number of people that are allowed to be creators of stuff as opposed to consumers of stuff on the internet.” He and the founders dare to suggest that Suno could attract a user base bigger than Spotify’s. If that prospect is hard to get your head around, that’s a good thing, Rodriguez says: It only means it’s “seemingly stupid” in the exact way that tends to attract him as an investor. “All of our great companies have that combination of excellent talent,” he says, “and then something that just seems stupid until it’s so obvious that it’s not stupid.”

Well before Suno’s arrival, musicians, producers, and songwriters were vocally concerned about AI’s business-shaking potential. “Music, as made by humans driven by extraordinary circumstances … those who have suffered and struggled to advance their craft, will have to contend with the wholesale automation of the very dear-bought art they have fought to achieve,” Reid writes. But Suno’s founders claim there’s little to fear, using the metaphor that people still read despite having the ability to write. “The way we think about this is we’re trying to get a billion people much more engaged with music than they are now,” Shulman says. “If people are much more into music, much more focused on creating, developing much more distinct tastes, this is obviously good for artists. The vision that we have of the future of music is one where it’s artist-friendly. We’re not trying to replace artists.”

Though Suno is hyperfocused only on reaching music fans who want to create songs for fun, it could still end up causing significant disruption along the way. In the short term, the segment of the market for human creators that seems most directly endangered is a lucrative one: songs created for ads and even TV shows. Lucas Keller, founder of the management firm Milk and Honey, notes that the market for placing well-known songs will remain unaffected. “But in terms of the rest of it, yeah, it could definitely put a dent in their business,” he says. “I think that ultimately, it allows a lot of ad agencies, film studios, networks, etc., to not have to go license stuff.”

In the absence of strict rules against AI-created content, there’s also the prospect of a world where users of models like Suno’s flood streaming services with their robo-creations by the millions. “Spotify may one day say ‘You can’t do that,’” Shulman says, noting that so far Suno users seem more interested in just texting their songs to a few friends.

Suno only has 12 or so employees right now, but they plan to expand, with a much larger permanent headquarters under construction on the top floor of the same building as their current temporary office. As we tour the still-unfinished floor, Schulman shows off an area that will become a full recording studio. Given what Suno can do, though, why do they even need it? “It’s mostly a listening room,” he acknowledges. “We want a good acoustic environment. But we all also enjoy making music — without AI.”

Suno’s biggest potential competitor so far seems to be Google’s Dream Track, which has obtained licenses that allow users to make their own songs using famous voices like Charlie Puth’s via a similar prompt-based interface. But Dream Track has only been released to a tiny test user base, and the samples released so far aren’t nearly as impressive-sounding as Suno’s, despite the famous voices attached. “I just don’t think that, like, making new Billy Joel songs is how people want to interact with music with the help of AI in the future,” Shulman says. “If I think about how we actually want people doing music in five years, it’s stuff that doesn’t exist. It’s the stuff that’s in their head.”

AI is just going to democratize the sh*t out of everything isn't it.

Chairman_Mao wrote:

AI is just going to democratize the sh*t out of everything isn't it.

AI tech bros are going to try to democratize everything they think they can make money off of, yes.

It's kind of the whole problem. A lot of these would be neat if they were a "we wanted to see if it was possible" kind of academic experiment, and once they knew it was they went and tried to create a commercial version with stuff they have the proper permissions and licenses for, but they are all just attempts to create and sell a "disruptive" start-up, and by the time the legal and ethical issues actually catch up to them, the people like Rodriguez have already cashed out and are on to the next project.

Nvidia is using AI to turn game characters into chatbots

Nvidia is showing off how developers have started using its AI “digital human” tools to voice, animate, and generate dialogue for video game characters. At the Game Developers Conference on Monday, the company released a clip of Covert Protocol, a playable tech demo that showcases how its AI tools can allow NPCs to respond in unique ways to player interactions, generating new responses that fit the live gameplay.

In the demo, players take on the role of a private detective, completing objectives based on conversations with AI-powered NPCs. Nvidia claims that each playthrough is “unique,” with players’ real-time interactions leading to different game outcomes. John Spitzer, Nvidia’s vice president of developer and performance technologies, says the company’s AI tech “may power the complex animations and conversational speech required to make digital interactions feel real.”

Covert Protocol was built in collaboration with Inworld AI, an AI gaming startup, and uses Nvidia’s Avatar Cloud Engine (ACE) technology — the same tech that powered the futuristic ramen shop demo that Nvidia released last May. The new Covert Protocol demo doesn’t show how effective these AI-powered NPCs are for real gameplay, instead showing a selection of clips of NPCs spitting out different voice lines. The line delivery and lip-syncing animations both feel robotic, as though an actual chatbot were talking at you through the screen.

Inworld says it’s planning to release Covert Protocol’s source code “in the near future” to encourage other developers to adopt Nvidia’s ACE digital human tech. Inworld also announced a partnership with Microsoft in November 2023 to help develop Xbox tools for creating AI-powered characters, stories, and quests.

Google has a new head of Search — and she’s all in on AI

There are few bigger jobs at Google than being the person who runs Search. As of today, there’s a new person in that seat: Liz Reid, who has been at Google for more than 20 years and has most recently been leading the company’s efforts with AI search, known as Search Generative Experience (SGE).

Reid’s promotion is part of a bigger change within Google’s Search team. Pandu Nayak, a longtime executive overseeing ranking and quality, is now going to be chief scientist of Search. He’ll be replaced by Cheenu Venkatachary, who has also been working on AI products in Search. Meanwhile, Cathy Edwards, who led a lot of the work on Google News and Google Discover, is taking a job on Google’s long-term bets team.

In some ways this is just a lot of normal corporate machinations: people getting new titles and new jobs and moving around after years in a specific role. And Prabhakar Raghavan, who has long overseen Search, ads, Assistant, and much more, is still in charge.

But let’s read some tea leaves anyway, shall we? The simplest way to look at these moves is as more evidence that Google believes AI is the future of search. For 25 years, we’ve all learned to type keywords into a search box and expect a bunch of ranked links in return. In an AI-powered, multisearch-based world, you might instead upload a photo, and the Gemini model could tell you what’s in it and how to buy it. You might speak a question into your headphones and get a fully formed answer out the speakers.

This is not surprising: CEO Sundar Pichai and others have been saying for years that language models and other AI systems can both improve the quality of search results and completely change the way we think about gathering information online. Now, as Google reckons both with a rapidly moving AI landscape and a web increasingly filled with AI-generated, SEO-ified junk, making Google Search good is both harder and more important than ever.

Reid, in particular, has spent the last couple of years working on both AI search and multisearch, which together point to a totally new way of thinking about Google. In a LinkedIn post announcing her new role, Reid noted all these new search tools, including the new Circle to Search feature, Google Lens, and other new ways of thinking about search. “With SGE,” she wrote, “we are able to serve a wider range of information needs and answer new types of questions, including more complex questions, like comparisons or longer queries.” And there’s more coming soon, she said.

Prederick wrote:

IMAGE(https://pbs.twimg.com/media/GJEQcukWMAAq9d4?format=jpg&name=large)

As someone who once cut pizzas in a pizza shop, this is blasphemy.

So does it count as man-made horrors beyond my comprehension or not, because the AI was made by mankind, but the image it produced was not.

While I haven’t heard this particular program, there’s a pretty good text-to-singing algorithm I heard a year or so ago that was really good and convincingly “human,” and there have been tons of software that can generate from scratch basically any genre of music for years now, it seems like this company just made something that does both rather than something new.

It doesn't do both, it just makes the music. It outsources generating the lyrics and title to ChatGPT.

Stengah wrote:

It doesn't do both, it just makes the music. It outsources generating the lyrics and title to ChatGPT.

It outsources the lyrics, yes, but the article says it generates both the singing and instrumentation, which was the “doing both” part.

The article makes it seem unprecedented but there’s a text-to-singing algorithm I’ve seen periodically advertised on facebook since last year that can do multiple different voices pretty competently, and music generation software predates the current “AI” bubble.

AI content flooding Facebook so thoroughly that Boomers can’t find their nazi rage bait among all the fake pizzas is legitimately the best use case for AI generated content that I could imagine.

At least until the AIs start generating nazi rage bait. Then we’re finished.