Month: August 2022

If you’re impressed by the recent spate of text to image generators, get ready for the next step in AI artistry: text-to-video.

While the huge compute costs and scarcity of text-to-video datasets have stunted the technique’s growth, recent research has brought the promise closer to reality.

A computer artist called Glenn Marshall has given a glimpse at the potential.

Greetings, humanoids

Subscribe to our newsletter now for a weekly recap of our favorite AI stories in your inbox.

The Belfast-based composer recently won the Jury Award at the Cannes Short Film Festival for his AI film The Crow.

Marshall had previously earned plaudits for an AI-generated Daft Punk video, but he applied a different approach to The Crow.

While his earlier technique turned text into random visual mutationsThe Crow uses an underlying film as an image reference.

“I had been heavily getting into the idea of ​​AI style transfer using video footage as a source,” Marshall told TNW.

“So every day I would be looking for something on YouTube or stock video sites, and trying to make an interesting video by abstracting it or transforming it into something different using my techniques.

“It was during this time I discovered Painted on YouTube — a short live action dance film — which would become the basis of The Crow.”

Marshall fed the video frames of Painted to CLIP, a neural network created by OpenAI.

He then prompted the system to generate a video of “a painting of a crow in a desolate landscape.”

My AI film ‘The Crow’ wins Jury Award at Cannes!https://t.co/WHDsI7UzJM pic.twitter.com/Ww1DGyBbxw

— Glenn Marshall (@GlennIsZen) August 24, 2022

Marshall says the outputs required little cherry-picking. He attributes this to the similarity between the prompt and underlying video, which depicts a dancer in a black shawl mimicking the movements of a crow.

“It’s this that makes the film work so well, as the AI ​​is trying to make every live action frame look like a painting with a crow in it, so I’m meeting it half way, and the film becomes kind of a battle between the human and the AI ​​— with all the suggestive symbolism.”

In the future, Marshall wants to add 3D animation to his AI creations. He’s also exploring CLIP-guided video generation, which can add detailed text-based directions, such as specific camera movements.

That could lead to entire feature films produced by text-to-video systems. Yet Marshall believes even his current techniques could attract mainstream recognition.

He says The Crow is now eligible for submission to the prestigious BAFTA Awards.

“I haven’t got a speech prepared, but I fantasize about collecting an award, in the role of a herald of AI, and proclaiming to the star-studded audience that [for] each and every one of you, actor, director, set designer, costume designer, artist, composer… AI is coming, and you’ll find yourself in a very different job soon — or out of a job all together.”

If you’ve been following the GPU market these past couple of years, it’s been a wild ride. But now that we’re at the tail end of a massive spike in GPU prices, Nvidia is set to release its next big flagship GPU, the RTX 4090.

Details and leaks about the upcoming 40-series have been all of the place, with some extreme rumors about power consumption and price. Either way, it’s likely to take a spot among the most powerful GPUs you can buy. We don’t know what all will end up being true, but here’s everything we’ve heard so far.

Kiklas/Shutterstock

release date

We won’t know any of the final details or pricing until Nvidia officially launches the RTX 4090. But we do know that Nvidia traditionally drops new graphics cards every two years in the fall, which makes sometime between mid-September and the end of October the most likely scenario.

The last time around was at the height of a global health and supply crisis. While things are definitely chaotic this year, chip supply chains, at least, are in much better shape than they were in 2020. This means the RTX 4090 could get into consumers’ hands much faster than the RTX 3090.

To add more support to an upcoming fall release, Nvidia leaked a testing schedule back in late May and included August and September for mass production of the unit.

Igor’s Lab

Of course, this could still be off. After all, a rumor back in May pegged the RTX 4090 release date for mid-July. Obviously, this didn’t happen, and it doesn’t fit with Nvidia’s customary fall release.

On the other hand, it is possible we could see Nvidia delay the event into early 2023. The reason is simple: They have too many RTX 3000-series chips on hand. Logistics bottlenecks during the pandemic meant chips piled up in warehouses, and Nvidia may hope to offload some more of these 2020 GPUs before dumping a new version onto the market.

Right now, rumors are suggesting that Nvidia will release only the RTX 4090 and hold off on the mid-range 4000-series cards while they offload surplus stock. This would help bring down pricing on older, but still capable. chips, and put them more in line with AMD chips.

Price

The current RTX 3090, arguably the king of graphics cards right now, dropped from a high of $1,999 and has settled around $1,499. The RTX 4090 will probably end up north of that number, although we can’t see it reaching $2,000 territory.

Pricing is important right now, especially with AMD snipping at Nvidia’s heels. The AMD Radeon RX 7000 series of GPUs promise to be beasts as well, but at a lower price. The current-gen AMD Radeon RX 6900 XT retails for $900, or $500 less than the current Nvidia RTX 3090 Ti. Nvidia’s recent financial woes in the gaming sphere could even prompt Nvidia to get more aggressive with pricing, but we’ll have to wait and see.

Architecture rumors

According to the rumors, the Nvidia RTX 4090 will most likely include 16,384 CUDA cores, 144 ray tracing cores, and 576 tensor cores. Potentially, this could give it a sizzling 40% performance increase over the RTX 3090.

It will also include 24GB of VRAM, the same as the 3090. However, Twitter leaker Kopite7kimi says the RTX 4090 will have a base clock speed of 2235 MHz and can boost up to 2520 MHz, more than double that of the RTX 3090.

Power draw

If rumors are to be believed, the RTX 4090 will be a power-chewing monster with an unquenchable appetite for electricity. Put more succinctly, it will require up to 600 watts of power when pushed to its maximum. With the way some are already downclocking their GPUs to keep heat and noise down, this amount of power could be a problem.

Although it will come with a standard 12-pin PCIe Gen 5 power connector, chances are there will need to be some kind of adapter in order to draw that kind of power. It could be built in or may need to be purchased separately. Nvidia usually doesn’t nickel-and-dime consumers, so we don’t see the latter being likely.

Still, 600 watts is a lot of power. The previous-gen RTX 3090 gulps up to 385 watts, which is a lot. The RTX 4090 will move enough energy to heat your home, meaning it’ll be beyond time to upgrade your power supply.

Jacob Roach / Digital Trends

According to tech watchdog Igor’s Lab, the RTX 4090 will need 450 watts just to power the GPU. The VRAM, fans, inductor, caps, and MOSFET will need the rest.

Because of this massive need for power, the RTX 4090 will get hot. We’re talking “surface of the sun” levels of heat — enough to melt the unit under sustained load. It would require a massive cooling unit to keep temperatures under control.

We reported on just such a thing back in June. The RTX 4090 has a three-slot BFGPU cooler much like the RTX 3090 Ti, but with more fins and more heat sink surface. It will likely keep the same dual flow-through as the RTX 3090, and the logo will light up LED lights.

performance

Power doesn’t necessarily translate into performance, but the RTX 4090 needs to find a way to convert that energy into pure graphical fulfillment. There’s no denying that high-end Nvidia GPUs are among the best in the world, even if they’re overpriced and hard to get. But the RTX 4090 could end up being on a different level altogether.

For starters, the card will supposedly be able to achieve a score of 19,000 on a Time Spy Extreme benchmark test, according to kopite7kimi. This would make it the fastest GPU ever built and nearly double the RTX 3090’s score of 10,000. Other rumors suggest the RTX 4090 can hit 4K 160 fps with both ray tracing and DLSS enabled. Gamers are going to love this card, if these rumors prove correct.

But more shocking is the raw performance. The RTX 4090 is set to become the first GPU to break the 100 teraflops bar. These are truly phenomenal numbers, but let’s be real: This is an enthusiast card only. Performance such as this sits on the edges of consumer needs.

Gamers will be able to play Cyberpunk 2077 or Elden Ring at full settings. The only problem could be what your electricity bill will look like later.

Editors’ Recommendations

Wikipedia has a problem. And Meta, the not-too-long-ago rebranded Facebook, may just have the answer.

Let’s back up. Wikipedia is one of the largest-scale collaborative projects in human history, with more than 100,000 volunteer human editors contributing to the construction and maintenance of a mind-bogglingly large, multi-language encyclopedia consisting of millions of articles. Upward of 17,000 new articles are added to Wikipedia each month, while tweaks and modifications are continuously made to its existing corpus of articles. The most popular Wiki articles have been edited thousands of times, reflecting the very latest research, insights, and up-to-the-minute information.

The challenge, of course, is accuracy. The very existence of Wikipedia is proof positive that large numbers of humans can come together to create something positive. But in order to be genuinely useful and not a sprawling graffiti wall of unsubstantiated claims, Wikipedia articles must be backed up by facts. This is where citations come in. The idea – and for the most part of this works very well – is that Wikipedia users and editors alike can confirm facts by adding or clicking hyperlinks that track statements back to their source.

Citation needed

Say, for example, I want to confirm the entry on President Barack Obama’s Wikipedia article stating that Obama traveled to Europe and then Kenya in 1988, where he met many of his paternal relatives for the first time. All I have to do is to look at the citations for the sentence and, sure enough, there are three separate book references that seem to confirm that the fact checks out.

By contrast, the phrase “citation needed” is probably the two most damning in all of Wikipedia, precisely because they suggest that there’s no evidence that the author didn’t conjure the words out of the digital ether. The words “citation needed” affixed to a Wikipedia claim is the equivalent of telling someone a fact while making finger quotes in the air.

Citations don’t tell us everything, though. If I were to tell you that, last year, I was the 23rd highest-earning tech journalist in the world and that I once gave up a lucrative modeling career to write articles for Digital Trends, it appears superficially plausible because there are hyperlinks to support my delusions.

The fact that the hyperlinks don’t support my alternative facts at all, but rather lead to unrelated pages on Digital Trends is only revealed when you click them. For the 99.9 percent of readers who have never met me, they might leave this article with a slew of false impressions, not the least of which is the surprisingly low barrier to entry into the world of modeling. In a hyperlinked world of information overload, in which we increasingly splash around in what Nicholas Carr refers to as “The Shallows,” the existence of citations themselves appear to be factual endorsements.

Meta wades in

But what if citations are added by Wikipedia editors, even if they don’t link to pages that actually support the claims? As an illustration, a recent Wikipedia article on Blackfeet Tribe member Joe Hipp described how Hipp was the first Native American boxer to challenge for the WBA World Heavyweight title and linked to what seemed to be an appropriate webpage. However, the webpage in question mentioned neither boxing nor Joe Hipp.

In the case of the Joe Hipp claim, the Wikipedia factoid was accurate, even if the citation was inappropriate. Nonetheless, it’s easy to see how this could be used, either deliberately or otherwise, to spread misinformation.

It’s here that Meta thinks that it’s come up with a way to help. Meta AI (that’s the AI ​​research and development research lab for the social media giant) has developed what it claims is the first machine learning model able to automatically scan hundreds of thousands of citations at once to check if they support the corresponding claims. While this would be far from the first bot Wikipedia uses, it could be among the most impressive — although it’s still currently in the research phase, and not in use on actual Wikipedia.

“I think we were driven by curiosity at the end of the day,” Fabio Petroni, research tech lead manager for the FAIR (Fundamental AI Research) team of Meta AI, told Digital Trends. “We wanted to see what was the limit of this technology. We were absolutely not sure if [this AI] could do anything meaningful in this context. No one had ever tried to do something similar [before].”

Understanding meaning

Trained using a dataset consisting of 4 million Wikipedia citations, Meta’s new tool is able to effectively analyze the information linked to a citation and then cross-reference it with the supporting evidence. And this isn’t just a straightforward text string comparison, either.

“There is a component like that, [looking at] the lexical similarity between the claim and the source, but that’s the easy case,” Petroni said. “With these models, what we have done is to build an index of all these webpages by chunking them into passages and providing an accurate representation for each passage … That is not representing word-by-word the passage, but the meaning of the passage . That means that two chunks of text with similar meanings will be represented in a very close position in the resulting n-dimensional space where all these passages are stored.”

xkcd

Just as impressive as the ability to spot fraudulent citations, however, is the tool’s potential for suggesting better references. Deployed as a production model, this tool could helpfully suggest references that would best illustrate a certain point. While Petroni balks at it being likened to a factual spellcheck, flagging errors and suggesting improvements, that’s an easy way to think about what it might do.

But as Petroni explains, there is still much more work to be done before it reaches this point. “What we have built is a proof of concept,” he said. “It’s not really usable at the moment. In order for this to be usable, you need to have a fresh index that indexes much more data than what we currently have. It needs to be constantly updated, with new information coming every day.”

This could, at least in theory, include not just text, but multimedia as well. Perhaps there’s a great authoritative documentary that’s available on YouTube the system could direct users toward. Maybe the answer to a particular claim is hidden in an image somewhere online.

A question of quality

There are other challenges, too. Notable in its absence, at least at present, is any attempt to independently grade the quality of sources cited. This is a thorny area in itself. As a simple illustration, would a brief, throwaway reference to a subject in, say, the New York Times prove a more suitable, high-quality citation than a more comprehensive, but less-renowned source? Should a mainstream publication rank more highly than a non-mainstream one?

Google’s trillion-dollar PageRank algorithm – certainly the most famous algorithm ever built around citations – had this built into its model by, in essence, equating a high-quality source with one that had a high number of incoming links. At present, Meta’s AI has nothing like this.

If this AI was to work as an effective tool, it would need to have something like that. As a very obvious example of why, imagine that one was to set out to “prove” the most egregious, reprehensible opinion for inclusion on a Wikipedia page. If the only evidence needed to confirm that something is true is whether similar sentiments could be found published elsewhere online, then virtually any claim could prove technically correct — no matter how wrong it might be.

“[One area we are interested in] is trying to model explicitly the trustworthiness of a source, the trustworthiness of a domain,” Petroni said. “I think Wikipedia already has a list of domains that are considered trustworthy, and domains that are considered not. But instead of having a fixed list, it would be nice if we can find a way to promote these algorithmically.”

Editors’ Recommendations

Nikita Gourianov, a physicist at Oxford university, yesterday published a scathing article full of wild, damning claims about the field of quantum computing and the scientists who work in it.

According to Gourianov, the quantum computing industry has been led astray by greedy physicists who’ve hyped up the tech’s possibilities in order to rip off VCs and get paid private-sector salaries for doing academic research.

Double, double

Per Gourianov’s article, the real problems started in the 2010s after investors started taking notice of the hype surrounding quantum physics:

As more money flowed in, the field grew, and it became progressively more tempting for scientists to oversell their results. With time, salesman-type figures, typically without any understanding of quantum physics, entered the field, taking senior positions in companies and focusing solely on generating fanfare. After a few years of this, a highly exaggerated perspective on the promise of quantum computing reached the mainstream, leading to a greed and misunderstanding taking hold and the formation of a classical bubble.

Greetings, humanoids

Subscribe to our newsletter now for a weekly recap of our favorite AI stories in your inbox.

Gourianov’s entire premise seems to hinge on their assertion that “despite years of effort nobody has yet come close to building a quantum machine that is actually capable of solving practical problems.”

They illustrate their argument by pointing out that Rigetti, IonQ, and D-Wave (three popular quantum computing companies) combined have failed to turn a sufficient profit.

According to Gourianov:

The reality is that none of these companies — or any other quantum computing firm, for that matter — are actually earning any real money. The little revenue they generate mostly comes from consulting missions aimed at teaching other companies about “how quantum computers will help their business”, as opposed to genuinely harnessing any advantages that quantum computers have over classical computers.

Finally, Gourianov’s conclusion leaves no doubt as to their feelings on the subject:

Well, when exactly the bubble will pop is difficult to say, but at some point the claims will be found out and the funding will dry up. I just hope that when the music stops and the bubble pops, the public will still listen to us physicists.

Toil and trouble

In the words of the great Jules Winnfield, Samuel Jackson’s character from the classic film Pulp Fiction, “Well, allow me to retort.”

I have but five words I’d like to say to Gourianov, and they are: IBM, Google, Amazon, Microsoft, and Intel.

I don’t think we need to do a deep dive into big tech’s balance sheets to explain that none of those companies are in any financial danger. Yet, each of them is developing quantum computers.

It’s unclear why Dr. Gourianov would leave big tech out of the argument entirely. There are dozens upon dozens of papers from Google and IBM alone demonstrating breakthrough after breakthrough in the field.

Gourianov’s primary argument against quantum computing appears, inexplicably, to be that they won’t be very useful for cracking quantum-resistant encryption. With respect, that’s like saying we shouldn’t develop surgical scalpels because they’re practically useless against chain mail armor.

Per Gourianov’s article:

Shor’s algorithm has been a godsend to the quantum industry, leading to untold amounts of funding from government security agencies all over the world. However, the commonly forgotten caveat here is that there are many alternative cryptographic schemes that are not vulnerable to quantum computers. It would be far from impossible to simply replace these vulnerable schemes with so-called “quantum-secure” ones.

This appears to suggest that Gourianov believes at least some physicists have pulled a bait-and-switch on governments and investors by convincing everyone that we need quantum computers for security.

This argument feels a bit juvenile and like a borderline conspiracy theory. Governments around the world have been working in tandem with experts from companies such as Google spinout SandboxAQ and IBM for several years to address the encryption issue.

No serious person involved in the decision-making is going to be confused about how math works because of crappy marketing hype or a misleading headline.

Gourianov’s rhetoric reaches a peak as they appear to accuse physicists of manipulating the hype around quantum computing out of sheer greed:

Some physicists believe, in private, that there is no problem here: why not take advantage of the situation while it lasts, and take the easy money from the not-too-sophisticated investors? Earning a private-sector level salary whilst doing essentially academic research is a pretty good deal, after all.

That’s quite the accusation.

Quantum computing bubble?

On the whole, however, it feels like Gourianov’s chief complaint isn’t that quantum computers don’t work, it’s that they aren’t very useful. dr Gorianov isn’t wrong. The technology is far from mature.

But make no mistake, today’s quantum computing systems do work. They just don’t work well enough to replace classical computers for many useful functions — yet.

Images: IBM

Looking at the above roadmap, keep in mind that IBM was founded in 1911. It didn’t build the IBM 5150, the company’s first PC, until 1981.

Along the way, a lot of reputable scientists said the PC market was a bubble. The naysayers claimed it was not only pointless for consumers, but that there were just too many problems to overcome in order to make computers affordable and useful for personal use.

We all know how that worked out for IBM. Do we need to even get into what Intel, Microsoft, Amazon, and Google have accomplished over the course of their ventures? They each have their own roadmaps concerning how they’re approaching the STEM challenges involved in quantum computing. So does MIT, Harvard, Oxford and myriad other universities.

I’m not a big tech shill by any means, but there’s a lot to be said about a fistful of companies worth somewhere around the one trillion dollars mark each deciding that a future-facing technology vertical is worth wagering their bankbooks on.

It’s beyond the scope of this article to address every non-trillion dollar company in the quantum computing field. But, having spoken to dozens of people working at various quantum computing companies, including the ones Gourianov mentioned, it’s clear to me that nobody building quantum computers has any misconceptions about their capabilities — not even the C-suite executives.

If VCs are confused and media hype is distorting the tech’s possibilities, I’d call that par for the course.

I can’t think of a single modern technology that mainstream journalism gets right all the time. And a significant portion of wealthy VCs are going to be both eager and ignorant about any given tech — shall we discuss AI or Web3 investments too?

The future is now

In my opinion, it would take a scientific shocker on par with discovering a viable antithesis for Newton’s laws for the bottom to fall out of the quantum computing industry. We’re not talking about a theoretical technology, we’re talking about a nascent one.

Quantum computers are here now. But like the IBM 5150 in 1981, they don’t really do anything that regular computers of their day can’t already do.

Still, I’d be interested in hearing what anyone who said the PC market was a bubble back in 1981 has to say about it now.

I imagine we’ll all see quantum computers differently in 40 years.

Perhaps quantum computing is a bubble market for VCs looking for short-term ROI projects, but the technology isn’t going anywhere.

There’s overwhelming evidence that today’s quantum computing technology is rapidly advancing to the point where it can help us solve problems that are infeasible for classical computation.

Maybe there are a bunch of greedy scientists out there peddling unwarranted optimism to VCs and entrepreneurs. But I’d wager that the curious scientists and engineers who chose this field because they actually want to build quantum computers outnumber them.