Tech Brew Ride Home - Tue. 01/09 – Do We Need ChatGPT In Our Cars?

Episode Date: January 9, 2024

OpenAI responds to the lawsuit from the NYTimes. Do you need ChatGPT in your car? VW thinks you do. Sony teases a “spatial” VR headset. Apple only wants you to call the Vision Pro “spatial computing.” And how you too can sign up to get a demo of the Vision Pro in a couple of weeks. Sponsors: Miro.com/podcast Links: OpenAI and journalism (OpenAI Blog) Amazon Debuts Video-Streaming Feature That Rivals Apple AirPlay (Bloomberg) Volkswagen says it’s putting ChatGPT in its cars for ‘enriching conversations’ (The Verge) Sony teased a ‘spatial’ VR headset with a smart control ring (The Verge) Instagram and Facebook Will Stop Treating Teens Like Adults (WSJ) Apple Vision Pro demos in retail stores will begin on February 2 (9to5Mac) Apple asks developers not to refer to their visionOS apps as ‘AR’ or ‘VR’ (9to5Mac) Luma raises $43M to build AI that crafts 3D models (TechCrunch) Hewlett Packard Enterprise Near Deal to Buy Juniper Networks (WSJ) CES Videos YouTube Channel Learn more about your ad choices. Visit megaphone.fm/adchoices

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 TechMeme right home for Tuesday, January 9th, 2024. I'm Brian McCullough today. OpenAI responds to the lawsuit from the New York Times. Do you need chat GPT in your car? VW thinks you do. Sony teases a spatial VR headset. Apple only wants you to call Vision Pro spatial computing and how you two can sign up to get a demo of the Vision Pro in a couple of weeks. Here's what you missed today in the world of tech. Open AI has responded to the New York Times's lawsuit over AI training data. Open AI. asserts training is fair use, and there is an opt-out that so-called regurgitation is a rare bug, and they make the claim that the New York Times, quote, manipulated its models to get them to,

quote, their past articles verbatim. Quote, while we disagree with the claims in the New York Times lawsuit, we view it as an opportunity to clarify our business, our intent, and how we build our technology. Our position can be summed up in these four points, which we flesh out below. One, we collaborate with news organizations and are creating new opportunities. Two, training is fair use, but we provide an opt-out because it's the right thing to do. Three, regurgitation is a rare bug that we are working to drive to zero. And four, the New York Times is not telling the full story. Training AI models using publicly available Internet material is fair use as supported by longstanding and widely accepted precedents. We view this principle as fair to creators, necessary for innovators, and critical for U.S. competitiveness. Memorization is a rare, failure of the learning process that we are continually making progress on, but it's more common when particular content appears more than once in training data, like if pieces of it appear on lots of different public websites. So we have measures in place to limit inadvertent memorization and prevent regurgitation in model outputs. We also expect our users to act responsibly,

intentionally manipulating our models to regurgitate is not an appropriate use of our technology and is against our terms of use. Interestingly, the regurgitations the New York Times induced appear to be from years old articles that have proliferated on multiple third-party websites. It seems they intentionally manipulated prompts, often including lengthy excerpts of articles in order to get our model to regurgitate. Even when using such prompts, our models don't typically behave the way the New York Times insinuates, which suggests they either instructed the model to regurgitate or cherry-picked their examples from many attempts. We regard the New York Times as a lawsuit to be without merit. Still, we are hopeful for a constructive partnership with the New York Times

and respect its long history, end quote. The Times itself didn't immediately respond to a request for comment, even from their own reporters writing about this. Amazon has announced Mattercasting, an Apple Airplay-like feature to send video from different platforms to Amazon devices, initially supporting Echo Show devices, quoting Bloomberg. The feature called Mattercasting is part of a push by Amazon to create interoperable services and alternative to the proprietary technology developed by Apple and Google. It will make it easier for iOS and Android phones to send video to Amazon devices, such as its fire TV boxes and sticks, as well as the Echo Show 15 smart display.

The new system unveiled Tuesday at the CES conference in Las Vegas will initially just send content from Amazon's Prime video app to Echo Show devices. Mattercasting will begin supporting fire TV technology within months, the company said. By later this year, the feature will work with a range of other video services, including Plex, Pluto TV, Sling TV, Stars, and ZDF, Amazon said, unquote. Apple and Google weren't part of this announcement, though they have worked together with Amazon on the matter standard. Amazon clearly hopes to move away from proprietary standards like Airplay and Chromecast to maybe level the device playing field just a bit. I suppose this was inevitable. Volkswagen plans to integrate chat GPT into its cars to control basic functions and answer what it says are general knowledge questions. This will start in Europe in Q2 of this year,

quoting the verge. The chatbot will be available across. VW's lineup, including in Tiguan, Passat, and Gulf, as well as the automaker's ID family of electric vehicles. The feature will come to Europe first and is being considered for customers in the U.S., though plans have yet to be finalized. VW is using ChatGBT-T to augment its IDA-in-car voice assistant to enable more naturalistic communication between car and driver. Vehicle owners can use the new super-powered voice assistant to control basic functions like heating and air conditioning or to answer, again, general knowledge questions. If you're scratch, your head, wondering why you would possibly need chat GPT in your car, VW says future functions may

help prove its worth. Quote, enriching conversations, clearing up questions, interacting in intuitive language, receiving vehicle-specific information, and much more, purely hands-free, the company says, end quote. VW said you invoke this by saying, hello, Ida, or a steering wheel button, without needing a new account or app installation. OpenAI doesn't get access to driving data, and VW claims robust data protection by instantly deleting queries, unlike basic vehicle voice assistants, which, if you think about it, are pretty simplistic because they don't need to do much. VW's enhanced system developed with Serence's automotive-grade chat GPT integrations offers advanced conversational capabilities and comprehensive response capabilities surpassing

common systems like Google's assistant and Amazon's Alexa. But again, to do what? I don't feel like I'll be having a personal conversation with my car anytime soon, but who knows. That was an announce from CES, of course, and at their keynote yesterday, Sony teased a spatial VR headset designed for 3D content creation with 4K OLED micro displays and powered by Snapdragon XR2 plus Gen 2, available later in 2024. So are they going after the Vision Pro and the quest, quoting the verge? The headset, which is powered by the Snapdragon XR plus Gen 2 platform and outfitted with 4K OLED displays isn't primarily for gamers like PlayStation VR. Instead, Sony's CEO Kenichiro Yoshida said that this new headset is made for professionals who work with 3D,

allowing wearers to expand the, quote, creation space by overlaying virtual objects into physical spaces. The demo showed users manipulating virtual objects while using two controllers, a ring and a pointer that fits into your hand. Meanwhile, the front part of the headset flips up to let users switch between virtual and physical spaces. Sony says its proprietary rendering technology allows for real-time, high-definition, and realistic rendering of textures of 3D objects and facial expressions of human characters. With seamless access to virtual objects, creators can work in real space with an immersive development experience, Yoshita said. Sony previously made a spatial reality display that allowed users to see digital objects in 3D using face and eye tracking.

That tech here is obviously a lot different, but it seems to be designed for the same type of pro-content creator. Sony says the new system will come out later this year, but there's no word on pricing yet, end quote. Meta has announced plans to start automatically restricting teen, Instagram, and Facebook accounts from harmful content, including posts about self-harm in the coming weeks. Quoting the journal, this marks the biggest change the tech giant has made to ensure younger users have a more age-appropriate experience on its social media sites. The new content restrictions come as more than 40 states are suing meta, alleging the tech company misled the public about the dangers its platforms post. to young people. Teen accounts, that is, accounts of under 18 users, based on the birth date entered during sign-up, will automatically be placed into the most restrictive content settings. Teens under 16 won't be shown sexually explicit content at all. On Instagram, this is called

sensitive content control while on Facebook. It is known as Reduce. Previously, teens could choose less stringent settings. Teen users can't opt out of these new settings. The new restricted status of teen accounts means teens won't be able to see or or search for harmful content, even if it's shared by a friend or someone they follow. For example, if a teen's friend had been posting about dieting, those posts will no longer be visible to the teen. However, teens might still see content related to a friend's recovery from an eating disorder. Teens won't necessarily know what they aren't seeing, a company's spokeswoman said, because

the content simply won't be available to them. Meta says it consulted with experts in adolescent development to determine what types of content are inappropriate for teens. Meta said its algorithms already avoid recommending harmful content to teens in its video reels and explore pages. With the new changes, such content will no longer be shown to teens in their feed and stories. The changes will be automatically applied to existing teen accounts starting this week. Newly created teen accounts will also be restricted to age-appropriate content, end quote. Okay, I'm here to answer a question that a lot of folks are going to be having real soon.

How do you get to demo the new Apple Vision Pro in an Apple store? The answer came in an email from Apple to those who signed up to be on the Vision Pro promo mailing list, it read, starting at 8 a.m. on Friday, February 2nd, we invite you to sign up for a demo of Apple Vision Pro at your local Apple store. Demo times will be available Friday through the weekend on a first-come, first-served basis. We can't wait to see you there, end quote. How long is the demo? What does the demo even consist of? We don't know. They didn't tell us, and they specifically said, first-come, first-serve, but only in relation to that first weekend, which suggests they will have more of a sign-up system at some point, or maybe that language

suggests that demos will only happen that first weekend, in which case, doubly mark your calendar and get there hours ahead of time. But I can't believe that'll be the case. Apple will want as many people as possible to get a sense of how this thing works at some point, won't they? This is maybe here nor there, or else it's interesting tea leaves to read. Apple has released new Vision OS developer guidelines asking developers to refer to Apple Vision Pro apps as spatial computing apps, not AR, VR, VR, XR, or MR apps. quoting 9 to 5 Mac. On a new page on the Apple developer website, the company details all the recommendations and requirements for launching Apple Vision Pro apps on the App Store. There are details

on how developers should test their apps, prepare screenshots, create icons, and more. Apple also provides details on how developers should name and describe their apps, while requirements such as using the Vision OS brand with the lowercase V and mentioning Apple Vision Pro without breaking up the full product name. Don't come as a surprise. There's one rather intriguing recommendation. The company is asking developers not to refer to VisionOS apps using terms such as AR, augmented reality, VR, Virtual Reality, XR, Extended Reality, or MR, mixed reality. Instead, Apple says that VisionOS apps are spatial computing apps. The request is somewhat contradictory, since Apple itself has been referring to Vision Pro as a product with augmented and virtual reality

technologies. It's true that most of Apple's marketing materials say that Vision Pro is a spatial computing platform. However, at WWDC 2024, Apple CEO Tim Cook unveiled the product as, end quote, entirely new AR platform, end quote. Of course, this is more of a marketing and branding decision than anything else by emphasizing that Vision Pro is a spatial computing platform. Apple wants to distinguish its product from any other ARVR headset out there. Internally, before the official announcement, Vision OS was referred to as XROS among Apple engineers. Apple also doesn't want to have Vision Pro associated with meta. as mixed reality products, which are known by the term Metaverse. In an interview in 2022, Apple's

vice president of global marketing, Greg Jawswiak said that Metaverse is a, quote, word I'll never use, end quote. Interesting raise. Luma, which is building AI models that generate 3D models from text prompts, has raised a $43 million series B. A source says at a valuation between $200 and $300 million, quoting TechCrunch. A couple of years ago, Alex U. and Emmett Jane came together to found a company that had let people capture objects in 3D using their smartphones no additional equipment required. At the time, Yu was an AI researcher at UC Berkeley, while Jane was an Apple employee building out the Vision Pro's multimedia experiences. Their company, Luma, launched a smartphone app in 2021, which quickly gained traction going on

to attract millions of users, just over 2 million as of publication time. But now, as generative AI tech floods the channels, you and Jane hope to evolve Luma into something bigger, and, with any luck better than they originally envisioned. Luma today announced that it'll begin leveraging a compute cluster of around 3,000 NVIDIA-100 GPUs to train new AI models that can, in use words, see and understand, show, and explain, and eventually interact with the world, end quote. The first phase of this plan entails creating models capable of generating 3D8 objects from text descriptions. Luma launched one such model on its Discord server earlier this year called Jeannie.

The next will be developing next-gen-gen-generative AI models that address what you characterizes as the uncanny valley problem in current gen AI, end quote. Finally today, here are two companies we rarely talk about, but it's a big enough potential deal that I think it's worth mentioning. Sources are telling the Wall Street Journal that HPEE is in advance talks to buy Juniper networks for about $13 billion in a deal that could be announced as soon as this week. Quoting the journal, Texas-based HPEC, as Hewlett-Packard Enterprise is known, is a cloud services provider to customers ranging from small businesses to large corporations

and governments. HPE is an offshoot of the iconic tech company founded in 1939 by William Hewitt Packard and David Packard. HPE shares closed Monday at $17.72, bringing its market capitalization to roughly $23 billion. Juniper's stock ended trading at $30.22, taking its market value to about $9.6 billion. A tech bubble-era darling, its shares have languished for years and currently trade at a fraction of what they once fetched. Sunnyvale, California-based juniper sells communications. network services and equipment like routers and switches to technology, telecommunications, financial, and other customers. Juniper also operates a growing artificial intelligence business known as Mist AI, which it says uses AI and machine learning to optimize users' experiences around wireless access. In 2015, HPE was split between an enterprise services producer and a computer and printer maker known as HP.

A deal between HPE and Juniper would be one of the biggest recent technology takeovers. Tech, normally the biggest engine of merger volume, has been in the doldrums amid a slowdown in dealmaking, hit particularly hard by tougher antitrust scrutiny out of Washington, end quote. Ah yes, AI, these days, whenever anything happens in tech, especially deals, just assume it's happening because of AI. As soon as I get done with a quick fund call early this afternoon, I'm going to throw up some more YouTube videos of cool things I've seen from CES, so look for that. that electric motorcycle thing from yesterday was interesting. Like, yeah, with all of the cameras and sensors cars have now to avoid collisions, motorcycles should be nothing but, you know,

LIDAR and sensors because they're more dangerous, right? Anyway, talk to you tomorrow.