Tech Brew Ride Home - Mon. 12/09 – OpenAI Releases Sora
Episode Date: December 9, 2024OpenAI’s Sora model is here! TikTok is still in trouble. China does more turnabouts is fair play, this time with Nvidia. I Apple, belatedly, getting serious about gaming on the Vision Pro? And did G...oogle just make a huge, historic leap forward in quantum computing? Sponsors: 1Password.com/ride MackWeldon.com promocode: brian Links: Sora is here (OpenAI) OpenAI’s controversial Sora is finally launching today. Will it truly disrupt Hollywood? (LATimes) OpenAI releases Sora, its buzzy AI video-generation tool (CNBC) Appeals Court Upholds U.S. Ban of TikTok (WSJ) Nvidia Hit With China Probe in Global Tech War Escalation (Bloomberg) Apple, Sony Discuss Teaming Up on Gaming Controllers for Vision Pro (Blooomberg) Quantum Computing Inches Closer to Reality After Another Google Breakthrough (NYTimes) Google Debuts New Quantum Chip, Error Correction Breakthrough, and Roadmap Details (HPCWire) Learn more about your ad choices. Visit megaphone.fm/adchoices
Transcript
Discussion (0)
On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco.
Hey, who did this to you?
What happened next turned the story into a political firestorm.
Reports have identified the victim as Bob Lee, the founder of Cash App.
From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16.
Welcome to the tech meme right home for Monday, December 9th, 2024. I'm Brian McCalla today. OpenAI's
SORA model is here. Tick-Tock is still in trouble. China does more turnabouts as fair play this time with
NVIDIA. Is Apple belatedly getting serious about gaming on the Vision Pro? And did Google just make a huge
historic leap forward in quantum computing? Here's what you miss today in the world of tech.
As long-rumored and much anticipated, OpenAI's video-generating AI model SORA is now available to use at
SORA.com. Users can generate videos up to 1080p resolution, up to 20 seconds long, in widescreen,
vertical, or square aspect ratios. Users can bring their own assets to extend remix and blend
or generate entirely new content from text. SORA is included as part of a plus account at no
additional cost on OpenAI, and you can generate up to 50 videos at 480p or slightly fewer at
720p per month. Note, it is not available in the EU or China.
Quoting from the OpenAI announcement.
For those who want more SORA, the pro plan includes 10x more usage, higher resolutions, and longer
durations.
We're working on tailoring pricing for different types of users, which we plan to make available
early next year.
The version of SORA we are deploying has many limitations.
It often generates unrealistic physics and struggles with complex actions over long durations.
Although SORA Turbo is much faster than the February preview, we're still working to
make the technology affordable for everyone.
We're introducing our video generation technology now to give society time.
to explore its possibilities and co-develop norms and safeguards that ensure it's used responsibly
as the field advances. All Sora generated videos come with C2PA metadata, which will identify a video
as coming from SORA to provide transparency and can be used to verify origin. While imperfect,
we've added safeguards like visible watermarks by default and built an internal search tool
that uses technical attributes of generations to help verify if content came from SORA, end quote.
Quoting the Los Angeles Times. The tool will be accessible.
for people 18 or older where chat GPT is available except for the United Kingdom,
Switzerland, and countries in the European Economic Area. OpenAI said it is working on enabling
Sora in those locations. The company is also preparing a free version of Sura. We really believe
that Sora can open doors for people to explore and share their creativity visually, especially
without extensive resources or training, said Suki Mansour, Sora Artist's program lead for Open AI in an
interview. As we know, filmmaking is very expensive. Sora is designed as a creative collaborator,
so the hope is that it helps artists bring very ambitious projects to life without expensive resources, Mansour said.
We think that this is raising the bar for what's possible with video creation.
And quoting CNBC, the AI video generating model works similarly to OpenAI's image-generating AI tool Dali.
A user types out a desired scene and SORA will return a high-definition video clip.
Sora can also generate video clips inspired by still images and extend existing videos or fill in missing frames, end quote.
Get ready for a bunch of videos.
flooding all the social channels in three, two, one. Something broke late Friday. That is an update to
something I was wondering about, so I want to put it on your radar. Remember that whole TikTok ban thing?
What's happening with that, given all the winds of political change about? Well, the journal was
reporting that TikTok still faces a U.S. ban if it is not sold by, and do the math on this,
January 19th. That is because a federal appeals court ruled that Congress does indeed have the
the power to take action to protect U.S. interests. I said do the math because inauguration day
is the next day, January 20th. So this is a signed law. It would go into effect because even if
the incoming administration wanted to prevent this from happening, they would need to sign a
whole new countervailing law, which they wouldn't have time to do. Quote, the decision relied
heavily on warnings from U.S. officials that the Chinese government can exert its will on the
app's parent company, Beijing-based bite dance, potentially giving it the ability to access
U.S. users' data and manipulate what they see on the platform. The decision by the U.S. Court of Appeals
for the District Court of Columbia Circuit rejected arguments by TikTok and several of its star users
that the ban was an unconstitutional infringement on free speech. The First Amendment exists to
protect free speech in the United States. Here, the government acted solely to protect that freedom
from a foreign adversary nation and to limit that adversary's ability to gather data on people
in the United States, Judge Douglas Ginsburg wrote for the court.
TikTok is expected to appeal to the Supreme Court, but the justices are under no obligation to hear the case.
The Supreme Court has an established historical record of protecting Americans' right to free speech,
and we expect they will do just that on this important constitutional issue.
A TikTok spokesman said,
Unfortunately, the TikTok ban was conceived and pushed through based on inaccurate, flawed, and hypothetical information,
resulting in outright censorship of the American people, end quote.
bite dance has said it can't and won't sell its U.S. operations. The Chinese government has
opposed a forced sale, preferring to keep TikTok's proprietary algorithm and source code under Chinese
control. President-elect Donald Trump's pending return to the White House adds a layer of uncertainty
to what happens next. Trump sought to restrict TikTok in his first term, but he muddied his position
earlier this year, expressing concern that a ban would shift users to rival Facebook. The ban doesn't
make it a crime for TikTok's 170 million U.S. users to keep using the app, but it prohibits mobile
app stores such as Google's and Apple's from letting users download or update it and bars internet
hosting services from supporting the app effectively shutting it down in the U.S.
Google and Apple haven't indicated how they would comply with the law. Apple didn't respond
to a request for comment, and Google, when reached, didn't have anything to add.
The statute subjects violators to fines of $5,000 multiplied by the number of people in the U.S.
who, quote, accessed, maintained, or updated the TikTok app.
app. That means, in theory, companies defying the ban could face civil penalties in the hundreds
of billions of dollars. TikTok will likely deploy a two-pronged approach as it tries to salvage its
U.S. presence. To postpone the ban's current January 19th effective date, the company would likely
need an emergency stay from the Supreme Court. The High Court has been sensitive to free speech
claims, but also traditionally has deferred to other branches on national security matters.
The app's other potential avenue is through Trump, who in theory could refuse to enforce the ban or
invoke provisions of the law that allow the president to lift the ban if his administration
determines the site is no longer under Chinese control, end quote. Quite possibly related to that,
according to Chinese state media, the state administration for market regulation opened an
antitrust monopoly investigation into invidia tied to its Melanox acquisition back in 2020.
So more tip for tat, I guess, quoting Bloomberg. The move against Nvidia is Beijing's latest
repost to escalating U.S. technology curves coming just a week after the Chinese government
banned exports of several materials with tech and military applications. InVIDIA's market value has
ballooned this year on demand for chips that can run artificial intelligence programs, making it one of
the most valuable publicly traded companies, and China's largest corporate target in the trade war so
far. China had approved NVIDIA's $7 billion acquisition of Melanox on condition that the
Israeli computer networking equipment maker provide information about new products to rivals
within 90 days of making them available to Nvidia.
France also targeted Nvidia in a probe looking at chips used in AI last year.
The company may face antitrust charges one day there.
Benoit Corre, the head of France's antitrust agency set out a press conference in July.
The European Union started a similar early stage probe in 2023 to examine alleged anti-competitive abuses in AI chips, end quote.
Mark German, Apple scoop Monday.
Might Apple finally be working on getting some games onto the Vision Pro?
Quoting Mark, among the many drawbacks of Apple's Vision Pro, its failure as a gaming device is perhaps the most glaring.
There are games on the device's app store, but there aren't many, if any, compelling titles.
Apple's attempt to make the Vision Pro into a gaming machine didn't extend much beyond supporting its existing arcade service
and allowing people to use PlayStation 5 and Xbox controllers.
This marks a stark contrast with meta's headset approach, which relies heavily on games to drive demand.
Apple should have made gaming a priority by creating its own hand controllers and building a service specifically for the device.
It also could have helped fund the development of big budget games, the kind of AAA-level titles that entice an audience.
But Apple made a decision during the Vision Pro's development that hurt its gaming prospects.
Urged on by designer Johnny Ive and hardware executive Mike Rockwell,
the company chose to focus on the device's novel interface, which relies on eye movements and hand gestures.
Bringing a controller accessory to the equation would have been confusing to develop.
developers and further increased costs. Without gaming, the Vision Pro's $3,500 price is a particularly hard
sell. Think about it. There are plenty of people willing to shell out thousands of dollars on gaming
rigs, but there aren't many who would spend that kind of cash for the Vision Pro's current features.
Since the Vision Pro went on sale in February, I'm told that the company has sold fewer than
half a million units, and a large number of Vision Pro buyers, those who haven't returned it,
aren't using the product as much as Apple anticipated, according to internal data gathered by the
company. There's hope, though. Apple is now working on a major
effort to support third-party hand controllers in the device's VisionOS software and has teamed up with
Sony Group to make it happen. Apple approached Sony earlier this year, and the duo agreed to work
together on launching support for the PlayStation VR2's hand controllers on the Vision Pro. Inside,
Sony, the work has been a month's long undertaking, I'm told, and Apple has discussed the plan
with third-party developers asking them if they'd integrate support into their games. This is
critical because the currently supported PS-5 and Xbox controllers are optimized for living-room
console's not virtual reality devices. To allow for the needed precision, Apple really needs
dual hand controllers that support six degrees of freedom, like the accessories that Meta and
Sony offer for their own headsets. Beyond gaming, these hand controllers could be used for productivity
tasks and media editing. Though the Vision Pro's hand and eye tracking works well, it's still hard
to make precise moves. If Apple wants to bring programs like Final Cut Pro and Adobe Photoshop into
this realm, hand controllers will be necessary. As for supporting the PlayStation VR2 controllers,
Apple and Sony originally aimed to announce this capability weeks ago, but the rollout has been
postponed. My expectation is that an announcement will still come at some point unless it gets
abruptly scrapped. One hiccup is that Sony doesn't currently sell its VR hand controllers as a standalone
accessory. The company would need to decouple the equipment from its own headset and kickoff
operations to produce and ship the accessory on its own. As part of the arrangement, Sony would sell
the controllers at Apple's online and retail stores which already offer PS5 versions. The move is meant
primarily for games on the Vision Pro, but the companies also created support for navigating the
device's operating system. The controller's thumbstick and directional pad could be used for
scrolling, while the trigger button could replace a finger pinch when clicking on an item, end quote.
This could end up being the headline of the year in tech that we look back on in history,
sort of like we look back on the attention is all you need paper for AI. Google says it had a major
quantum error correction breakthrough as it unveiled its Willow Quillow
quantum chip with 105 improved qubits, quoting the New York Times. Google said its quantum computer
based on a computer chip called Willow needed less than five minutes to perform a mathematical
calculation that one of the world's most powerful supercomputers could not complete in 10 septillion
years, that's 10 to the 24 years, a length of time that exceeds the age of the known universe.
Quantum computing, the result of decades of research into a type of physics called quantum
mechanics is still an experimental technology, but Google's achievement shows that scientists are
steadily improving techniques that could allow quantum computing to live up to the enormous
expectations that have surrounded this big idea for decades. When quantum computing was
originally envisioned, many people including many leaders in the field, felt it would never be
a practical thing, said Mikhail Lukin, a professor of physics at Harvard, and co-founder
of the quantum computing startup Q era. What has happened over the last year shows that it is no longer
science fiction, end quote. And quoting HPCWire. It's rare to catch Google in a talkative mood,
but the technology giant held a media analyst briefing before today's announcement with a handful
of prominent members of the Google Quantum AI team. Demonstrating the ability to break the QEC
threshold was really the biggest news of the day. Doing that required many advances in the
chip itself as Google progressed from its sycamore line to the will align. High error rates,
particularly in superconducting cubits, have limited their usefulness so far.
Newman, Google Research Scientists explained quantum information is extremely fragile. It can be disrupted by many things ranging from microscopic material defects to ionizing radiation to cosmic rays, and unfortunately to enact these large-scale quantum algorithms that we have all this promise. We need to sometimes reliably manipulate this information for billions, if not trillions of steps, whereas typically we see failures on the order of 1 in 1,000 or 1 in 10,000, so we need our qubits to be almost perfect, and we can't get there with engineering alone.
The way to make almost perfect qubits is with quantum error correction or QEC.
So the basic idea is you take many physical cubits and have them work together to represent a single logical cubit.
The physical cubits work together to correct errors, and the hope is that as you make these collections larger and larger, there's more and more error correction.
The problem is that as these things are getting larger, there's also more opportunities for error.
One of the researchers said, quote, in terms of go-to-market models, there is a number out there.
One of the most obvious in our consideration is offering access through cloud.
That is something we are currently working on.
Looking beyond that, I think there are certainly opportunities to support even jobs running at Google,
as well as potentially in larger, very specific on-prem or other types of models.
So TBD on those, but we are currently working on cloud.
Another Google researcher quickly added,
The Willow chip is a nice piece of hardware, and it has enormous raw compute power,
so we have ambitions to show that already with such a platform,
you may be able to do useful algorithms.
That's not the conversation for this press roundtable, but we take an IOU and watch us over the next year, end quote.
You know that thing where one by one, everybody in your house gets sick, catches a cold, but somehow you do not.
And everybody recovers one by one and you think, phew, somehow I made it.
And then, of course, you catch the cold.
It me.
I don't know if you can hear it in my voice.
I slept 12 solid hours last night, six to six.
If Open AI would hurry up and release SORA, I can do that segment.
Get this out to you and then go back to bed.
Talk to you tomorrow.
