Tech Brew Ride Home - Fri. 09/22 – Microsoft… Wins?

Starting point is 00:00:00 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 Friday, September 22nd, 2023. I'm Brian McCullough today. The UK CMA looks like it's folding so the Microsoft Activision Acquisition can go through now, question mark. YouTube unveils some cool new AI tools. Amazon is adding ads to prime video and raising the price.

Starting point is 00:00:54 And of course, the weekend long-range suggestions. Here's what you miss today in the world of tech. Looks like this might happen after all. The UK CMA has said that, Microsoft revised offer to acquire Activision, which includes selling some gaming rights to Ubisoft, appears to address the concerns that it had. Quoting Bloomberg, Microsoft's $69 billion acquisition of Activision Blizzard looks set to clear its final regulatory hurdle after the UK competition authorities signaled they will accept the latest concessions, ending a weight of more than a year and a half to complete the biggest ever gaming deal. The competition and markets Authority said Friday that Microsoft's proposal opens the door to the deal being cleared.

Starting point is 00:01:38 The restructured offer to sell some gaming rights to French publisher Ubisoft is expected to keep competition in cloud gaming open for years, the regulator said. It will consult on the offer until October 6th. The move represents a stunning turnaround for a deal that was once thought over after running up against concerns from antitrust regulators, including an initial veto from the UK in the spring. It gained unexpected momentum after Microsoft beat the Federal Trade Commission's Court challenge over the deal. The European Union cleared the deal with behavioral remedies in May. That left the CMA as the remaining regulatory hurdle. This is a new and substantially different deal, which keeps the cloud distribution of these important games in the hands of a strong

Starting point is 00:02:17 independent supplier, Ubisoft, rather than under the control of Microsoft, said Colin Rafferty, a senior director of mergers at the CMA. The CMA had previously vetoed the deal saying it could result in higher prices, fewer choices, and less innovation for UK gamers. The new offer means Microsoft can't limit access to Activision's key content to its own cloud gaming service or withhold those games from rivals, the regulator said. The parties have arguably gone further in the UK to secure regulatory clearance than elsewhere, said Alex Hafner, a competition lawyer at UK law firm, Flaggate. Still, the UK watchdog's turnaround allowed the CMA to climb down from an isolated position in regard to the other two regulators, said Jonathan Compton, a dispute resolution and

Starting point is 00:02:57 antitrust lawyer at DMH Stollard. In effect, the substance of the structural change is that Microsoft will not buy the rights to the cloud gaming rights owned by Activision, he said. This is, with respect, a fig leaf, end quote. I mentioned yesterday that YouTube also had an event this week, so I wanted to tell you what happened there, because to a larger degree, it plays into the whole AI narrative we've been covering extensively. YouTube unveiled YouTube Studio AI features that are coming in 2024 to let select creators test aloud, a dubbing tool that creates AI-generated dubs in other languages.

Starting point is 00:03:37 In other words, make a YouTube video once, but be able to distribute it into dozens of languages with the click of a button. Now that would be AI at scale, quoting TechCrunch. YouTube will introduce an AI dubbing tool called A Loud, which will be integrated into YouTube Studio. The tool only takes one click to get an AI-generated dub in another language, which the creator can then review before adding it to their video. This tool is testing with select creators now, and will open up more broadly next year, end quote. YouTube also announced Dreamscreen a shorts feature that lets users create an AI-generated video or image background using text prompts.

Starting point is 00:04:15 This is rolling out in early 2024, quoting TechCrunch again. For example, explain YouTube CEO Neil Mohan at the company's live event made on YouTube this morning. You could type in something as crazy as a panda drinking coffee and then the video image appears on the screen. The company also suggested other examples like underwater castles or imagery that you could have imagined in a dream like dragons or sci-fi moonrises. Mohan said he believes the technology would allow more people to publish on YouTube without feeling like they have to have a deep understanding of YouTube analytics or a full production studio. The Shorts platform today is now averaging over 70 billion daily views up from 50 billion in January, and YouTube expects AI will increase those numbers further.

Starting point is 00:04:57 At YouTube, we want to make it easier for anyone to feel like they can create, and we believe generative AI will make that possible, said Mohan, end quote. Now, there has been a lot of of energy around the margins in terms of AI for social media. Snap has been in this game for a while, of course. There are various startups like Cup of Soup that have gotten some attention. But remember, YouTube shorts is YouTube's answer to TikTok. So I wonder when we will see TikTok go hard at AI for both creation and consumption. Wow, another one of these. Apple has released emergency updates for iOS, iPad OS, and MacOS and watchOS to put three new zero-day vulnerabilities out of commission. This brings a total of 16-0-day patches released by Apple in

Starting point is 00:05:47 23, quoting bleeping computer. Two bugs were found in the WebKit browser engine and the security framework, enabling attackers to bypass signature validation using malicious apps or gain arbitrary code execution via maliciously crafted web pages. The third one was found in the kernel framework, which provides APIs and support for kernel extensions and kernel resident devices. Local attackers can exploit this flaw to escalate privileges. All three zero days were found and reported by Bill Markzak of the Citizen Lab at the University of Toronto's Monk School and Maddie Stone of Google's Threat Analysis Group. While Apple has yet to provide additional details regarding the flaws's exploitation in the wild, citizen lab and Google threat analysis group, security researchers have often disclosed zero-day bugs abused in targeted spyware attacks,

Starting point is 00:06:31 targeting high-risk individuals, including journalists, opposition politicians, and dissidents, end quote. Another domino has fallen. Amazon plans to introduce limited ad, ads to prime video in the U.S., the UK, Germany, and Canada early in 2024. The existing ad-free tier will now cost $2.99 a month extra, at least in the U.S., quoting Bloomberg. Ad-supported streaming will be the default on prime video in the U.S., UK, Germany, and Canada, starting early next year, the company said in a statement on Friday. The company has long offered video streaming as part of a package that also includes speedy shipping, music and other perks. Amazon said Prime subscribers will continue to pay $139 annually in the U.S.

Starting point is 00:07:19 but will be able to pay an additional $299 a month to avoid ads. Pricing in other countries will be announced later, the company said. In recent months, streaming services including Netflix and Walt Disney have introduced ads. Both Netflix and Disney have also raised subscription prices, while ads risk turning off viewers, services see them as a way to offset surging production costs amid rising competition. Amazon said it would aim to have fewer ads than on linear television and other streaming providers. Chief Executive Officer Andy Jassy has been scrutinizing how much the company spends on original TV programming. Bloomberg has reported in 2022, Amazon spent $7 billion on original shows, licensed programs and sports, up from $5 billion a year before.

Starting point is 00:08:01 Only Netflix and Disney spend more on streaming. Amazon shift towards ads leaves Apple TV plus as the only major streaming platform not to have them. them, end quote. Time for the weekend long read suggestions. First up, let's stick with Amazon. I mentioned that their event this week was somewhat muted. They still announced dozens of products, but not dozens and dozens like in years past, but with Panos-Penay jumping ship to ostensibly help lead development and design of these sorts of products, maybe we're not seeing a retrenching on Amazon's part so much as a change in strategy, because there's still one overriding vision that they've probably got their eye on. Remember all.

Starting point is 00:08:44 all the talk of them mapping your home, quoting the verge. Amazon is reportedly working on a projector that would turn any surface in your home into a screen. People familiar with a matter told Reuters that the projector is designed to do things like beam a recipe onto the wall above a stove and make Zoom calls that track you as you move around the house. A touch-interactive projected screen is an obvious move for the ambient smart home that Amazon is long proposed, one where the tech appears when you need it and disappears when you don't. I mean, that's literally the definition of a projector. The potential here for more applications around the home is quite exciting, as others have shown.

Starting point is 00:09:19 Appliance manufacturer Bosch showed off a kitchen projector at CES 2019 that used its MEMs-based projection tech with similar functions to Amazon's rumor device. I saw a demo of it showing you how to cut a cake precisely, but there are many more potential use cases, especially in a kitchen or bathroom with plenty of counterspace. The combination of mapping capabilities and a touchscreen projector opens up even more intriguing possibilities. A screen that could appear in front of you when you need to interact with it, wherever you are in your house, would be really useful. It could also cut down on the need for multiple interfaces in a smart home. We know that Amazon is working on mapping your home.

Starting point is 00:09:53 It just announced a new MapView UI for Smart Home Control in its Alexa app, which is coming to its Echo Hub's smart home controller next year. The obvious move here would be to use the Maps-I-Robots-Rumba's build of your home, which could be why Amazon purchased the vacuum company, however, that sale has stalled due to regulatory scrutiny. In its continued quest to make Alexa as profitable as its other service-driven revenue streams, a new piece of high-tech hardware to sell at close to cost is an odd play. However, since day one, the stick and Alexa-enabled gadget in every corner of customers' homes has been the company's approach, and based on this week's hardware event, that doesn't seem to be changing. A voice-enabled projector would

Starting point is 00:10:31 certainly open up more corners, albeit flat ones, end quote. Then, Quantum Magazine, turned me on to a new, different way to do generative AI. Although, is it generative? Maybe it's different. What if physics can provide a new model for doing things like this? Quote, Tegmark and his colleagues are learning whether other physics-inspired generative models might work as well as diffusion-based models or even better. Late last year, Tegmark's team introduced a promising new method of producing images called the Poisson Flow generative model, or PFGM. In it, data is represented by charge particles which combine to create an electric field whose properties depend on the distribution of the charges at any given moment. It's called a Poisson flow model because the movement

Starting point is 00:11:15 of charges is governed by the Poisson equation, which derives from the principle stating that the electrostatic force between two charges varies inversely with the square of the distance between them, similar to the formulation of Newtonian gravity. That physical process is at the heart of PFGM. Our model can be characterized almost completely by the strength and direction of the electric field at every point in space, said Yulin-Zu, a graduate student at MIT and co-author of the paper. What the neural network learns during the training process is how to estimate that electric field, and in doing so, it can learn to create images because an image in this model can be succinctly described by an electric field. PFGM can create images of the same quality as those produced

Starting point is 00:11:54 by diffusion-based approaches, and do so 10 to 20 times faster. It utilizes a physical construct, the electric field in a way that we've never seen before, said Hennel Hazan, a computer computer scientist at Tufts University, that opens the door to the possibility of other physical phenomena being harnessed to improve our neural networks. Diffusion and Poisson flow models have a lot in common besides being based on equations imported from physics. During training, a diffusion model designed for image generation typically starts with a picture, a dog, let's say, and then adds visual noise altering each pixel in a random way until its features become thoroughly shrouded, though not completely eliminated.

Starting point is 00:12:30 The model then attempts to reverse the process and generate a dog that's close to the original. Once trained, the model can successfully create dogs and other imagery starting from a seemingly blank canvas. Poisson flow models operate in much the same way. During training, there's a forward process which involves adding noise incrementally to a once sharp image and a reverse process in which the model attempts to remove that noise step by step until the initial version is mostly recovered. As with diffusion-based generation, the system eventually learns to make the images it never saw in training, end quote. This is just a public service. Last weekend, the internet was taken over by AI-generated, geometric art crafted using stable diffusion and control net, a neural network structure that allows

Starting point is 00:13:09 extra conditions to diffusion models. Well, the site, NFT, now, gives us a quick rundown about what this was all about. Quote, AI artist Kaliuga commented, honestly, I've seen a lot of AI-generated art. I've been in this field for a long time, and this is one of the most amazing works I've ever seen. You have done an outstanding job. It wasn't long before. Recognized names like Beeple and Jack Butcher offered their spin on the trend underlying its massive appeal. Paul Graham, co-founder of White Combinator has been a leading voice in the tech community for years. Graham's remark, this marked the moment when AI-generated art, metaphorically speaking, passed the Turing test in my eyes, he said, adding a layer of authority to the ongoing

Starting point is 00:13:47 discourse on AI art. Many may be wondering how exactly the art piece was created in the first place. The secret sauce is ControlNet, originally introduced in a paper titled, adding conditional control to text-to-image diffusion models. ControlNet enhances the stable diffusion process by adding deeper guidance-based, on extracted information from source images. This means artists, or rather artist AI collaborators, can replicate specific shapes, patterns, or subjects from an image with increased precision. In simple terms, while stable diffusion creates the general image, ControlNet fine-tunes the details helping bring the artist's vision to life.

Starting point is 00:14:21 Uyla shared that he will soon do a more detailed walkthrough of his process for the spiral artwork, but did share a bit on his workflow behind the checkered village, which used a similar approach, end quote. And finally today, not tech, but from Lapham's Quarterly, quick excerpt of a book that I Need to Buy Toot Sweet, How the English Language evolved in North America from colonial times until today. Quote, corn offers an example of how English words evolved in America. Before 1492, the plant that Americans call corn, Zia Mays, was unknown in England. The word corn was a general term for grain, usually referring to whichever cereal crop was most abundant in the region. For instance, corn meant wheat in England, but usually referred to oats in Ireland. When American corn came to Britain, it was named

Starting point is 00:15:06 Mays, the English version of mahis, an indigenous arachan word adopted by the Spanish. When the first colonists encountered it in North America, however, they almost always referred to it as corn or Indian corn, probably because it was the main cereal crop of the area. Americans repurposed other English words as well. For example, bug, which meant a bedbug in England, broadened to cover any insect, and sick, which referred specifically to a digestive upset, became a general term for any illness. What the British called timber, Americans called lumber. In England, lumber is old, discarded furniture and other items of the sort usually found in attics. Americans called a shop, a store, as in grocery store, perhaps from an archaic use of store to mean an abundant supply, and said fall for autumn.

Starting point is 00:15:50 Fall was short for fall of leaf, once a common phrase in England, but becoming obsolete by the 18th century. Americans also said, mad for angry, another English usage that diet. out in the old country, end quote. Guess what? At long last, we have a fresh new bonus episode for you this weekend, first in a while. It's a portfolio profile episode, and developers, listen up, you're going to want to hear about this one. It's called Play.

Starting point is 00:16:26 It's a whole new way to design and develop. You'll see why it's so cool that the Right Home Fund invested. Also, if you ever thought of doing a startup to target developers and designers, lots of interesting learnings in this one. Enjoy. Talk to you on Monday. Thank you.

Tech Brew Ride Home - Fri. 09/22 – Microsoft… Wins?

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.