Tech Brew Ride Home - Tue. 06/27 – What If OpenAI Did Its Own Copilot?

Episode Date: June 27, 2023

Checking in with the FTC vs. Microsoft case. Checking in with the effort to recover funds from FTX. What if OpenAI did its own Copilot? Would Microsoft be pissed? And reviews of the Pixel Fold include one reviewer’s unit dying after a mere few days. Sponsors: stpp.co/techmeme ZocDoc.com/techmeme Links: Microsoft exec was ready to ‘go spend Sony out of business’ to strengthen Xbox (The Verge) FTX Bankruptcy Team Says the Exchange Owed Customers $8.7B (CoinDesk) Thomson Reuters buys Casetext, an AI legal tech startup, for $650M in cash (TechCrunch) OpenAI Plans ChatGPT ‘Personal Assistant for Work,’ Setting Up Microsoft Rivalry (The Information) RIP to my Pixel Fold: Dead after four days (ArsTechnica) Google Pixel Fold review: closing the gap (The Verge) 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, June 27th, 2023. I'm Brian McCullough today. Checking in with the FTC versus Microsoft case. Checking in with the effort to recover funds from FTX.

What if OpenAI did its own co-pilot? Would Microsoft be pissed? And reviews of the Pixel Fold include one reviewer's unit dying after a mere few days. Here's what you missed today in the world of tech. Hey, did you know that whole FTC versus Microsoft hearing about whether or not the Activision acquisition can actually. go forward, is going on right now? How's that going? Well, it apparently uncovered a 2019 email by Xbox Game Studios Matt Booty encouraging Microsoft to buy content to, quote, spend Sony out of business,

end quote. Microsoft argues the company didn't actually pursue that as a strategy, but quoting the verge, we, Microsoft, are in a very unique position to be able to go spend Sony out of business, said Booty in a December 2019 email, referencing spending $2 billion or $3 billion in 2020 to avoid competitors getting ahead in content at a later date. Quote, it is practically impossible for anyone to start a new video streaming service at scale at this point, said Booty, referencing competitors like Google, Amazon, and Sony. Booty described content as a moat and that only Sony could really compete with Xbox GamePass. Quote, in games, Google is three to four years away from being able to have a studio up and running.

Amazon has shown no ability to execute on game content. Content is the one moat that we have in terms of a catalog that runs on current devices and capability to create new. Sony is really the only other player who could compete with GamePass, and we have a two-year and 10 million subs lead, end quote. Microsoft argues the email is old and that it never pursued such a strategy anyway, quote, this email is three and a half years old and predates the announcement of our acquisition by 25 months, says David Cuddy, general manager of public affairs at Microsoft, and a statement to the verge. It refers to industry trends we never pursued and is unrelated to the acquisition, end quote. The internal exchange was also sealed as part of a separate so-called gamers lawsuit recently.

The email shows Microsoft strategy and thinking around game content for its Xbox GamePass subscription in 2019. Microsoft has since acquired Bethesda for $7.5 billion and is trying to get its $68.7 billion proposed Activision Blizzard deal over the line. Both are a lot bigger than the two to three billion dollar figure booty floated in 2019. Microsoft also seriously considered acquiring Sega and Bungi with Xbox chief Phil Spencer, even going as far as requesting strategy approval from Microsoft's CEO Satcha Nadella to approach Sega Sammy regarding a potential acquisition of its Sega gaming studios. Both Sega and Bungi targets were part of a larger watch list that Microsoft have put together to acquire key studios and mobile developers to bolster Xbox GamePass, end quote.

Since we're doing check-ins, let's also check in on the FTX bankruptcy thing. How's that going? It's going okay, maybe? FTC's bankruptcy team says the exchange owed its customers around $8.7 billion after commingling and misusing their deposits, but so far has recovered around $7 billion in liquid assets. So that's not bad? Quoting CoinDesk. About $7.7.7.5.5. About 7,000. $6.4 billion of the money, the FTCS.com Exchange owed its customers was, quote, in the form of fiat currency and stable coin that had been misappropriated, according to the report filed on Monday. About $7 billion in liquid assets have been recovered so far, and those searching the company's assets, quote, anticipate additional recoveries.

A product of months of analysis and forensic auditing, the new report paints a picture of a company management and at least one senior lawyer knowingly misusing customer money, saying they, quote, lied to banks and auditors, executed false documents, and moved the FTC group from jurisdiction to jurisdiction, taking flight from the United States to Hong Kong to the Bahamas, in a continual effort to enable and avoid detection of their wrongdoing, end quote. The 33-page review is the second filed by John J. Ray III, the CEO who is trying to recover money for creditors, after he detailed an initial examination in April that provided a number of revelations of improper activity under founder and former CEO Sam Bankman-Fried's watch.

Bankman-Freed is facing a number of criminal charges set for an October trial in New York. The company is now in the midst of bankruptcy proceedings in Delaware. Ray has been trying to settle the exchanges affairs since its November collapse, and there have been some hints that its operations could be restarted as FTX 2.0. Those trying to trace FtX transactions and funding are finding the task, quote, extraordinarily challenging the report indicated, end quote. Thompson Reuters has agreed to acquire YC-backed case text, which builds AI-powered workflows and tools for lawyers for $650 million in cash,

expected to close in the second half of this year. So this maybe fits that whole M&A deluge kicking in for this AI moment that I spoke about yesterday. Tools for lawyers, tools for doctors, tools for, you name it. Incomments in any niche out there are probably racing to add AI to what they already do, or they could acquire someone who's actually gone and built it. Though in this case, case text is from a prior AI moment, quoting TechCrunch. Founded back in in 2013, case text, which TechCrunch has covered numerous times throughout its history, initially focused on creating both a community for attorneys to share knowledge and a service to give users free access to legal texts annotated by lawyers. But the company later pivoted,

embracing AI and machine learning to build automated workflows and tools for legal teams. Case Tech's flagship product is co-counsel, which taps AI to review documents, help with legal research memos, prepare, depositions, and analyze contracts. Case text was one of the few granted early access to OpenAI's GPT4 language model, which serves as the infrastructural back end for co-counsel. According to a press release case text, which has 104 employees, has a customer base exceeding 10,000 law firms and corporate legal departments, leading up to the acquisition, the company raised over $64 million from Union Square Ventures and others. For Thompson Reuters, the acquisition is part of a long-term strategy to embed

generative AI into its major business verticals. Legal, tax, accounting, and new. The company recently announced that it plans to spend some $100 million a year on AI and incorporate generative AI into its products in the second half of this year, and to set aside $10 billion for mergers and acquisitions, many AI-focused from now until 2025. The acquisition of case text is another step in our build, partner, and buy strategy to bring generative AI solutions to our customers. Thompson Reuters CEO Stephen Hasker said in a canned statement, we believe that case text will accelerate and expand our market potential for these offerings,

revolutionizing the way professionals work and the work they do, end quote. So as outlined by that last segment on case text being built on top of OpenAI's tech, folks are wondering, is OpenAI and GPTN number whatever just going to be like the underlying infrastructure for this new AI moment for people to build companies and products on top of, or could OpenAI do that itself? product ties itself in a more top-of-stack sort of way. Well, interesting. Sources are telling the information that Sam Altman has told developers in London about plans to create a, quote, super smart personal assistant for work, which is interesting for many reasons, not the least of

which being such a move would put OpenAI in direct competition with Microsoft's efforts, quote. With built-in knowledge about an individual and their workplace, such an assistant could carry out tasks such as drafting emails or documents in that person's style and with up-to-date information about their business. The assistant features could put OpenAI on a collision course with Microsoft, its primary business partner, investor, and cloud provider, as well as with other OpenAI software customers such as Salesforce. Those firms also want to use OpenAI software to build AI co-pilots for people to use at work. But for OpenAI building new chat GPT capabilities will be the focus of its commercial efforts according to Altman's comments and to other people with knowledge of

the company's plans. Companies are still in the first innings of making money from the latest crop of AI services, and the race is on to figure out what products and business models will create the most value. Large language models that allow ChatGBTGBT and other software to understand conversational commands are relatively new, although Microsoft is already charging a 40% premium to Office 365 customers that want to use OpenAIs, LLMs, to automate tasks such as creating PowerPoint presentations based on text documents, summarizing meetings, or drafting email responses. After the breakout success of ChatGPT last fall, OpenAI started charging individuals and companies for an upgraded version, which recently had more than 2 million subscribers and is on

pace to generate hundreds of millions of dollars per year, said a person with knowledge of the matter. To boost ChatGPT's value, OpenAI is considering opening a type of app store in which its customers would sell customized versions of the chatbot that could help other businesses draft marketing materials, identify fraudulent transactions, or build customer service chatbots, the information reported last week. The company has also been hedging its bets by planning to launch a new open source LLM to keep pace with open source releases by rivals such as meta platforms. For open AI, creating a reliable personal assistant would require a tremendous amount of work, according to artificial intelligence practitioners, including people who help open AI customers

use LLMs. Right now, customers access chat GPT through a web or mobile app, and all of its computations happen in the cloud, specifically through Microsoft's Azure, which uses thousands of specialized Nvidia server chips to power the chatbot. But to operate in a more personalized fashion and respond quickly the way Apple's Siri does, the prospective OpenAI assistant software may need to be partly stored on users' devices. That could take time because of difficulties in making the AI models small enough without also sacrificing the quality of their performance, according to some AI practitioners. The company would have to find a way to pare down ChatGPT's big software code base into smaller pieces and get permission from users to train the software on personal

data, such as in individual's emails, contacts, and information stored in business apps, such as Word and Google Docs. Microsoft, Google, and other business app makers typically let outside developers tap into such data if customers explicitly allow it, so OpenAI theoretically could get in that way. When Altman last month told some software developers in London about the plan to turn chat GPT into a personal assistant for work, he didn't give many specifics about what features it would have. Some are obvious, such as drafting emails or responding to messages using a person's unique style, but people at the company have also been watching how developers

have used the chatbot to create autonomous AI agents, set a person with knowledge of the matter. When the agents are given a goal to increase sales of a software product, for instance, they come up with distinct tasks they can carry out to achieve it, such as creating an app or an email template for outreach to potential customers. It wouldn't be surprising if OpenAI tried to incorporate such capabilities in a chat GPT assistant, end quote. Finally today, the pixel fold reviews are out, and there has been one high-profile issue. A pixel fold reviewer says the device's flexible OLED screen died after four days of light use,

starting at the bottom and moving upwards, likely due to debris getting inside, quoting Ron Amadeo at Ars Technica. A flame that burns twice as bright, burns half as long. That was my brief experience with the pixel fold, which was a wonderful little device until the display died, along with my hopes and dreams. I barely used it, but it was beautiful. I didn't do anything to deserve this. The phone sat on my desk while I wrote about it, and I would occasionally stop to poke the screen or take a screenshot or open and close it. It was never dropped or exposed to a significant amount of grit, nor had it gone through the years

of normal wear and tear that phones are expected to survive. This was the lightest possible usage of a phone, and it still broke. The flexible OLED screen died after four days. The bottom 10 pixels of the pixel fold went dead first, forming a white line of 100% brightness pixels that blazed across the bottom of the screen. The entire left half of the foldable display stopped responding to touch to, and an hour later, a white gradient started growing upwards across the display. Samsung, B-O-E, and pretty much every other company making foldable screens build these flexible OLEDs the same way. The OLED panel is covered in an ultra-thin glass. That's thin and flexible enough to survive the folding process, though it's not very durable, because the glass

can't stand up to the slightest bit of damage. The whole display is covered in a protective plastic layer. This essentially kills the firm, slippery glass surface we're all used to, but the interior glass layer provides some much-needed structure to what otherwise would be a very squishy plastic. This plastic layer is critical to the OLED survival, but it doesn't stretch to the edges. Every company that builds these screens leaves a margin around the perimeter of the display where there's no plastic layer, just a raw exposed OLED panel peeking out into the world. We would normally expect a foldable to break along the crease, where the screen sees the most stress, but mine died due to this exposed OLED gap. The tiniest bit of something got in there, and when I closed the display, the pressure

of the other display side was enough to puncture the OLED panel. I didn't see or feel anything when closing the device, but the display pixels started freaking out. After going over the device with a magnifying glass, I think I found where the puncture was. The exposed strip of the OLED panel is sandwiched between the edge of the screen protector and the raised bezel surrounding the phone. So even if you take care to wipe the display off, that exposed OLED perimeter acts as a gutter for any debris that lands on the phone. Even as I look at the dead flickering foldable now, it's easy to spot lint and other junk trying to accumulate in the OLED death zone. That appears to be what killed the phone, as I can see a near-microscopic nick mark near where the display first started having problems. Another problem may be the pixel's nearly flush bezels. We usually discuss the width of devices bezels, but here we're talking about their height.

the bezels on some foldables have a bit of height to them, so when the screen closes, there is still a small gap between the two halves of the display to protect against a small bit of debris being crushed between the halves. The pixel folds, bezels are almost flush with the screen protector, so when you close the phone, the two display halves almost touch. If you stick a wet piece of paper in the pixel fold and close it, both sides of the display will get wet. When I close the device with just a tiny speck of something in the gutter, that was enough to kill off the display, end quote. might all speak to why this morning, Google announced plans to offer do-it-yourself repairs for the pixel fold via its iFixit partnership, including the battery and interfolding display. Aside from the phone breaking, though, what have people thought of the fold? As ever, with reviews, I turn to the verge where Allison Johnson says, the landscape first design is great for videos, and the outer display feels natural, but it's heavy, it's pricey, and has limited multitasking features. This is from her conclusion, quote. With the Razor Plus, I kept uncovering new use cases where I felt the phone was truly saving me time and effort or letting me do something I couldn't do with a traditional

phone. While the pixel fold impressed me in many ways, I didn't get that same sense of I can't believe I'm getting away with this. Watching YouTube with the phone propped open while I eat lunch, gaming on a big screen, cross-referencing Chrome and Maps on the same screen as I plan out my day. These are all unequivocally excellent experiences on the pixel fold, but it all left me, with an impression of, that was nice, rather than I must have this in my life. The pixel folds heaviness plays a big part in this. It's uncomfortable to use it folded for very long, even if the outer screen format feels more natural than the remote control-esque galaxy Z-fold four. You notice its weight in your pocket, your tote bag, your purse, however you carry it

on your person. If I felt like I was getting an incredible experience in exchange for carrying the extra weight around, I wouldn't mind it as much, but that's not the case. First-generation foldables are tough. The pixel fold is far from disastrous, which you couldn't say about Samsung or Motorola's first foldable attempts, but those companies are several generations in now and have ironed out a lot of the kinks. Google seems to have learned from some of those early failures, too, but the pixel fold feels like it's still at least one generation away from realizing its potential. It's a delightfully gadgety gadget, and it's probably best suited for someone who wouldn't mind the weight and would utilize a big screen for on-the-go entertainment consumption often.

but if that sounds like more of a nice to have as it was to me, then I think you're better off waiting to see what the next generation has to offer, end quote. One more time, nothing for you today. Talk to you tomorrow.