Tech Brew Ride Home - Tue 11/05 – Apple: Maybe Smartglasses Are The Way To Go

Episode Date: November 5, 2024

Big hacking arrest in Canada. More evidence Apple is thinking: “You know, maybe smartglasses are the way to go…” More signs OpenAI is focusing on hardware. More signs they’re about to go for-profit. And what it’s actually like to use GPT Search? Is it a Google killer or no? Links: Canada Arrests Man Suspected of Hacks of Snowflake Customers (Bloomberg) Apple Explores Push Into Smart Glasses With ‘Atlas’ User Study (Bloomberg) Meta’s former hardware lead for Orion is joining OpenAI (TechCrunch) OpenAI in Regulator Talks to Become For-Profit Company (Bloomberg) Anthropic hikes the price of its Haiku model (TechCrunch) Uber’s Real Threat Isn’t From Robots (WSJ) ChatGPT Search is not OpenAI’s ‘Google killer’ yet (TechCrunch) 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 Tech meme right home for Tuesday, November 5th, 2024. I'm Brian McCullough today. Big hacking arrest in Canada. More evidence Apple is thinking, you know, maybe smart glasses are the way to go. More signs OpenAI is focusing on hardware. More signs they're about to go for-profit and what it's actually like to use GPT search. Is it a Google killer or no? Here's what you miss today in the world of tech. Bloomberg says Canadian authorities have arrested Alexander Connor Mauka, who is allegedly behind. in the June and July 2024 hacks of up to 165 snowflake users, including AT&T, quoting Bloomberg. Following a request from the U.S., Alexander Connor Malka was taken into custody

on a provisional arrest warrant on October 30th, according to Canada's Department of Justice. He is due to appear in court on Tuesday. The charges against Malka weren't immediately available, quote, as extradition requests are considered confidential state-to-state communications, we cannot comment further on this case, said Ian McLeod, spokesperson for Canada's Department of Justice. However, two people familiar with the hacks who asked not to be named so they could discuss confidential matters have identified Malka as the person behind the snowflake-related hacks. In addition, Austin Larson, senior threat analyst at the cybersecurity firm Mandient, alleged in a statement Monday evening, Alexander Connor Malka has

proven to be one of the most consequential threat actors of 2024, end quote. Malka launched a campaign in April against more than 100 organizations, leaving them, quote, reeling from significant data loss and extortion attempts, Larson said. He added that it highlighted the alarming scale of harm a single individual can cause using off-the-shelf tools, end quote. Companies including AT&T, Live Nation, and Advanced Auto Parts disclosed that they'd been affected by the attacks in June and July. In some cases, the hacker, or hackers, as it is not clear if others were involved, attempted to extort the companies by threatening to sell the data on criminal forums if they didn't pay up, according to cybersecurity analysts at Alphabet's Google.

The attacks resulted in the theft of millions of people's personal data. The hacker used stolen credentials that were available in places like cybercriminal forums to access customer accounts, which lacked security measures, such as multi-factor authentication, Snowflake has said, end quote. Welp, this is kind of what I was speculating on yesterday. This is quoting Mark German. Apple is exploring a push into smart glasses with an internal study of product currently on the market, setting the stage for the company to follow meta platforms into an increasingly popular category. The initiative, codenamed Atlas, got underway last week and involves

gathering feedback from Apple employees on smart glasses, according to people with knowledge of the matter. Additional focus groups are planned for the near future, said the people who ask not to be identified because the work is secret. The studies are being led by Apple's product systems quality team, part of the hardware engineering division. When Apple is considering whether to enter a new category, it often hosts secret focus groups to understand what people like about existing products. The company typically relies on employees rather than customers so it can avoid making its plans public. A representative for Apple declined to comment. Last month, Bloomberg News reported on another internal study. The company tested an app for people

with pre-diabetes that tracks their blood sugar changes and diet. The latest study suggests that Apple is moving forward with its own work on smart glasses. Bloomberg previously reported that the iPhone maker has been considering a foray into the market potentially challenging Meta's Rayban device, though an actual product would still be years away. The research will likely guide Apple in what features to include in its own glasses and help identify the ways the technology could be used. For years, Apple had sought to make a lighter-weight augmented reality version of the device, something that could be worn all day and potentially replace the iPhone. Work on that project has sputtered due to numerous technical challenges. Meta, meanwhile, found success with a more stripped-down

formula. It's $299 glasses created in partnership with Luxottica Group. Aren't true AR spectacles. They don't overlay information on the glass, but they let users shoot video, take phone calls, and ask questions to an AI assistant, end quote. Speaking of Caitlin Kalanowski, who served as Meta's head of AR glasses from March 22 to July 2024 and oversaw Orion, is joining Open AI to lead robotics and consumer hardware development. eyes emoji, quoting TechCrunch. Kalinowski is a hardware executive who began leading Meta's AR Glasses team in March 2022. She oversaw the creation of Orion, the impressive augmented reality prototype that Meta recently showed off at its annual Connect conference.

Kalinowski also led the hardware team behind Meta's virtual reality goggles for roughly nine years. Before that, she worked at Apple designing the hardware for MacBooks. Possibly Kalanowski will work with her old boss, former Apple executive Johnny Ive, on a new AI hardware device that open AI and Ives' startup love from are building together. In September, I've confirmed he was building a hardware product with OpenAI, describing it as a product that uses AI to create a computing experience that is less socially disruptive than the iPhone. OpenAI also recently started hiring research engineers for a robotics team that's aimed at helping OpenAI's partners incorporate its multimodal AI into their hardware. The reboot of OpenAI's robotics team comes roughly four years

after the startup disbanded its hardware research to focus its efforts on software. In 2018, OpenAI built a robot hand that could learn how to grip objects all on its own. Several companies already are incorporating OpenAI's models into their hardware. The most obvious is Apple, which will launch its chat GPT integration for the iPhone later this year. Another is the robotics company figure whose humanoid zero one robot leverages OpenAI's software for natural speech conversations, end quote. Meanwhile, Bloomberg says Open AI is in early talks with the California Attorney General's Office to become a for-profit company. A letter says OpenAI also discussed it with Delaware's AG. Quote, in 2019, OpenAI created a capped for-profit subsidiary to help fund the high

cost of AI model development. In 2023, OpenAI's chief executive officer, Sam Altman, was fired and rehired by its former nonprofit board. Altman's ouster followed tensions with the board overbalancing AI safety with the pressure to commercialize OpenAI software, among other issues. Unlike many other nonprofits, OpenAI holds incredibly valuable intellectual property in the form of its proprietary chat-GPT chatbot and related artificial intelligence technology. In California, the company has opened a dialogue with the Office of Attorney General Rob Bonta and will submit the details of its restructuring plan after the proposal is finalized, according to a person who declined to be identified because the discussions are private.

A spokeswoman for Banta's office said in a statement, it is committed to protecting charitable assets for their intended purpose without commenting on any discussions with OpenAI. The company plans to change to a public benefit corporation, which Bloomberg previously reported. The move will allow it to maintain its mission for social good, while operating as a for-profit business, Open AI chief strategy officer Jason Kwan told employees during a staff meeting in late September,

according to a person familiar with the matter. Kwan told employees, this new structure will preserve a non-profit arm that would own a material amount of the for-profit entity, said the person who declined to be identified. What stake the nonprofit will receive in the for-profit and how OpenAI's assets are valued will be key factors in regulatory approvals for the restructuring, according to legal experts. It's not as simple as just turning off your nonprofit status, said Darren Schaever, a San Francisco-based partner at Hansen Bridget LLP. Whatever value in those assets has to be properly accounted for, end quote. The process in California, which would involve

going back and forth with Banta's office typically can take a couple of months for an ordinary non-profit, Shaver said, but because California law requires whatever value is assigned to the nonprofit assets to be distributed to a charitable cause, and OpenAI's top asset is its intellectual property, the review could be complicated and drawn out. It's about convincing the Attorney General effectively that the assets are going to the right place, Shaver said. Delaware State Attorney General Kathleen Jennings asked OpenAI in an October 9 letter to submit its conversion plans once they are worked out for review by lawyers in her offices, Fraud and Consumer Protection Division. Open AI's conversion would also require a following up with the Secretaries of State in Delaware and

California on certain procedures as well as state and federal tax authorities, end quote. Claude 3.5 Haiku is here, but people aren't going to be happy. That's because it is priced at $1 per million input tokens, up 4x from Claude 3.0 Haiku's price. It also lacks image analysis features, quoting TechCrunch. Clod 3.5 Hiku, which Anthropic announced last month, matches or best the performance of Anthropics flagship model, Claude 3 opus on specific benchmarks. Available through Anthropics API and a number of third-party platforms, including AWS Bedrock. Claude 3.5 Hiku is useful for coding suggestions, data extraction, and labeling, and content moderation, Anthropics says, end quote. The expectation was that Anthropic would maintain its

predecessor's pricing, but the new model comes in with a significant cost increase. Users will now pay $1 per million input tokens and $5 per million output tokens four times more than Claude 3 Haiku's rates of $0.25 respectively. Additionally, as mentioned, unlike its predecessor, Claude 3.5 haiku launches without image analysis capabilities. According to Anthropics' head of developer relations, Alex Albert, the original Claude 3 Haiku will remain available for those prioritizing cost-effectiveness and image processing features. On the pricing front, Anthropics said this on X. During final testing,

Haiku surpassed Claude III Opus, our previous flagship model on many benchmarks, at a fraction of the cost. As a result, we've increased pricing for Claude 3.5 Haiku to reflect its increase in intelligence, end quote. You'll recall that I recently underlined the fact that Uber has become meaningfully profitable. Stock has soared in recent years, which is why I dug around for more about how their business is actually doing and found this from the journal. Turns out, the earnings recently weren't as great as I thought. Quote, its stock had been under pressure for much of the year because of worries about the prospect of Tesla launching a fleet of self-driving cabs and a network to run them on. The ride-hailing pioneers stock jumped nearly 11 percent the day after the event, as investors

thought Tesla's Robo Taxi Day was a bit of a dud. And then came Uber's own third quarter report last week, which included disappointing growth in its gross bookings and a projection for a further slowdown in the fourth quarter. In a twist the night before, DoorDash announced a deal to add Lyft services to its Dash Pass membership program, basically a team-up of Uber's two largest rivals in the U.S. market. Uber's stock has shed about 15% from its high point after the Tesla event, with last week's earnings report sparking the stock's single worst day in two years. That has put the shares back where they started a month ago, merely in line with the S&P 500's performance, and just behind that of the NASDAQ composite for the year.

The stock has badly trailed delivery rivals DoorDash and Instacart. There's always a wall of worry, wrote Bernstein analyst Nikiel Devnani in a report on Uber after last week's results. Wall Street isn't panicking, though. 85% of analysts covering the stock rate Uber as a buy, according to fact-set. Rising insurance cost played a role in the weaker bookings as prices for Uber rides rose to compensate and sparked in the words of Chief Executive Darakosha. kind of the typical elasticity from consumers who ride a bit less when prices go up.

The partnership between DoorDash and Lyft will likely make the Dash Pass membership program more competitive with Uber 1, though analysts are mixed on the ultimate impact. Devnani of Bernstein called the deal a negative for Uber in our minds as it creates a more competitive end market and undercuts Uber's unique content offered under Uber 1. Schwedda Kajuria of Wolf Research doesn't see a major impact on Uber, though. If you're a dash pass member, you probably don't have Uber 1, she said in an interview. And the risk from Tesla's theoretical Robotaxy Fleet remains as it was, nothing much. A lot remains to be seen about the electric vehicle maker's ability to complete and truly deliver self-driving technology.

Meanwhile, Uber's new partner, Waymo, keeps gaining traction. The Robotaxy operator owned by Google Parent Alphabet has managed to improve the wait times for its cars enough to close the gap with Uber and Lyft in San Francisco, according to a report by Evercore ISI on Monday. and Waymo will launch their first joint service in the Atlanta and Austin, Texas markets next year. We still believe there will be multiple AV winners, including both Waymo and Uber, with Uber serving as the leading demand aggregator for AVs, wrote Evercore analyst Mark Mahaney, referring to autonomous vehicles. Uber is now trading at less than 33 times forward earnings, a record low multiple that is a fraction of what Tesla now commands, according to fact set, end quote.

Finally today, a bit of a review. TechCrunch says chat GPT search is impressive to use, but it's unreliable for short queries containing just a few words, meaning it's far from being a Google killer. Quoting Maxwell Zeph. After using chat GPT search as my default search engine, you can too with OpenAI's extension, for roughly a day, I quickly switch back to Google. OpenAI search product was impressive in some ways and offered a glimpse of what an AI search interface could one day look like, but for now it's still too impractical to use as my daily driver. Chat GPT search was occasionally useful for surfacing real-time answers to questions that I would have otherwise had to dig through many ads and search optimized articles to find. Like others in this AI search category like Perplexity and U.com, it presents concise answers in a nice format. You get links to the information sources on the right side.

The headlines and short snippets give you a quick reality check on the AI-generated text. However, it often just felt impractical for everyday use. In its current form, chat GPT search is unreliable. for what people use Google for the most. Short navigational queries. Queries shorter than four words represent the bulk of searches on Google. These are often just a few keywords. They get you to the right web page, often one the user knows, but doesn't want to bother typing out. They're the kind of searches most people are barely even conscious they're making all day, and it's what Google tends to do very well. I'm talking about Celtics score, cotton socks, library hours, San Francisco weather,

cafes near me, and other queries that make Google the front page to the internet for billions of people. My test run with chat GPT search was quite frustrating at times, and it made me conscious of just how many keyword searches I perform in a day. I couldn't reliably find information using short queries, and for the first time in years, I actually longed for Google search. Don't get me wrong, Google has declined in quality for the last decade or so, largely because it's been flooded with ads, SEO, and questionable AI summaries. Still, I kept opening Google in a separate window during my test because chat GPT search couldn't get me a correct answer or web page. To its credit, chat GPT search is rather good at answering long-written-out search questions, something like

what American professional sports league has the most diversity, isn't a question you could easily answer but chat GPT search is pretty good at scraping multiple websites and getting you a decent answer in just a couple of seconds. Perplexity is also pretty good at these questions, and its search product has been around for well over a year. The problem is that most searches on Google are not such long questions. To really replace Google, OpenAI needs to improve these more practical short searches people are already making throughout their day, end quote. No more tech news for you today. Of course, there's a lot more news happening all day. Thankfully, not my Ballywick. Talk to you tomorrow.