Tech Brew Ride Home - Thu. 05/11 – Twitter Encrypts DMs? (Corrected)

Episode Date: May 11, 2023

Twitter finally rolled out encrypted DMs… or did they? More drips and drabs from yesterday’s Google event including an AI music generating tool that sounds pretty wild. Even the crypto miners are pivoting to AI. And why tech has warmed up to the concept of nearshoring. Sponsors: Grammarly.com/go Bloomberg.com/careers Links: Twitter's encrypted DMs are here — but only for verified users (Engadget) Join the waitlist for Google's generative AI tools, including search, Project Tailwind, & MusicLM (XDA Developers) Google makes its text-to-music AI public (TechCrunch) Google Cloud announces new A3 supercomputer VMs built to power LLMs (TechCrunch) AI Needs Specialized Processors. Crypto Miners Say They Have Them (Bloomberg) ‘Nearshoring’ Push Is Fueling Tech Job Demand in Latin America (Bloomberg) Learn more about your ad choices. Visit megaphone.fm/adchoices

Welcome to the Tech meme right home for Thursday, May 11th, 2020. I'm Brian McCullough today. Twitter finally rolled out encrypted DMs, or did they? More drips and drabs from yesterday's Google event, including an AI music generating tool that sounds pretty wild. Even the crypto miners are pivoting to AI now, and why tech has warmed up to the concept of nearshoring. Here's what you missed today in the world of tech. From the, they finally did it, but maybe not the way you wanted file, Twitter has, has rolled out encrypted DMs. But both sender and recipient must be verified users. There's no support for group messages, and message metadata is not encrypted. In fact, there's kind of a question of

what encryption we're actually talking about here, quoting in Gadget. It doesn't support group messages or any kind of media other than links, and the level of encryption appears to be less secure than what other apps offer. For one, message metadata is not encrypted. Furthermore, Twitter notes that, quote, currently, we do not offer protection against man in the middle attacks and suggests that the company itself is still able to access encrypted DMs without the participants knowing. If someone, for example, a malicious insider or Twitter itself as a result of a compulsory legal process, were to compromise an encrypted conversation neither the sender or the receiver would know, the company explains on a help page. It added that it's working on improvements that would make such exploits more, quote, difficult.

That's particularly notable because it falls far short. short of the standard Twitter owner Elon Musk has described when expressing his desire to add encryption for Twitter DMs. He has said he wants it to be impossible for the company to access users encrypted messages, even if, quote, someone puts a gun to our heads, end quote. In a tweet, Twitter security engineer Christopher Stanley acknowledged the shortcoming, quote, we're not quite there yet, but we're working on it. For those who are verified and want to try out the feature anyway, encrypted messaging can be accessed via the info menu. That's the same menu you use. to block or report a conversation within a particular DM. Once encryption is enabled, the encrypted

messages will appear as a separate message thread with labels at the top of the chat to indicate that the conversation is encrypted, end quote. Some bits and pieces to pick up now from yesterday's Google event. Remember how I told you some of the new features announced require you to sign up to get special access? Well, Google actually launched a dedicated labs page where users can sign up to test Google's early ideas for features and products, including search and workspace AI tools, quoting XDA developers. The Google Labs website went live a little over an hour after the Google I.O. 2023 event began and allows users to join waitlists for Google's experimental features. These include Search Labs, Google Workspace with AI, Project Tailwind, and Music LM.

Google says that these experiments have limited availability at the moment, and you'll need to join the waitlist for each experiment individually. To join some of the waitlists and experience the experimental features like Google Search Labs and Google Workspace with AI, you'll need to access the lab site with the Google Chrome browser. The Google Workspace Labs experiment lets you test out generative AI tools and services like Gmail, Google Docs, and Google Sheets. Though artificial intelligence suggestions have been built into Gmail for a while, generative AI can write entire emails on your behalf. In Google Docs and Google Sheets, it can create documents and spreadsheets based on the information provided. In a demo, Google asks Google Workspace Labs to make a

dog-walking business spreadsheet, and it automatically organized and populated a template spreadsheet. To sign up for the wait list, you'll need to be 18 years or older, and you will receive an email when it's time to try out Google Workspace Labs. Other current features in Google Labs include Project Tailwind and MusicLM. Project Tailwind is an AI-first notebook that can process inputted notes and respond to queries related to those notes naturally. In essence, it's an advanced version of a document search. MusicLM creates music based on what you're looking for, so asking for calming sounds might play relaxing music, end quote. Actually, let's go a bit more in depth on that music thing. MusicLM is an experimental AI tool that can turn text prompts into several song versions, and it's inside Google's AI Test Kitchen app on the web, Android, and iOS. coding TechCrunch. MusicLM lets users type in a prompt like soulful jazz for a dinner party or

create an industrial techno sound that is hypnotic and have the tool create several versions of the song. Users can specify instruments like electronic or classical as well as the vibe, mood, or emotion they're aiming for as they refine their music LM generated creations. When Google previewed MusicLM in an academic paper in January, it said that it had no immediate plans to release it. The co-authors of the paper noted that many ethical challenges, posed by a system like MusicLM exists, including a tendency to incorporate copyrighted material from training data into the generated songs. But in the intervening months, Google says it's been working with musicians and hosting workshops to, quote, see how the technology can empower the

creative process. One of the outcomes, the version of MusicLM in AI Test Kitchen, won't generate music with specific artists or vocals. Make of that what you will. It seems unlikely in any case that the broader challenges around generative music will be easily remedied. In 2020, Jay-Z's record label filed copyright strikes against a YouTube channel, vocal synthesis, for using AI to create JZ covers of songs like Billy Joel's We Didn't Start the Fire. After initially removing the videos, YouTube reinstated them, finding the takedown requests were, quote, incomplete. But deep-faked music still stands on murky legal ground. A white paper authored by Eric Sunray, now a legal intern at the Music Publishers Association,

argues that AI music generators, like Music L.M, violate music copyright by creating, quote, tapestries of coherent audio from the works they ingest in training, thereby infringing the United States Copyright Act's reproduction right. Indeed, AI like Music L.M learns from existing music to produce similar effects, as alluded to in the paper, a fact with which not all artists are comfortable. It might not be long before there's some clarity on the matter. Several lawsuits making their way through the courts will likely have a bearing on music generating AI, including one pertaining to the rights of artists whose work is used to train AI systems without their knowledge or consent. Time will tell, end quote.

One more real quick, because it might be useful for you to know this. Google Cloud also announced a new A3 GPU supercomputer VM powered by NVIDIA's H-100 GPUs built to deliver what Google calls the highest performance training for today's ML workloads, quoting TechCrunch. As we've seen LLMs and generative AI come screaming into our consciousness in recent months, it's clear that these models take enormous amounts of compute power to train and run. Recognizing this, Google Cloud announced a new A3 supercomputer virtual machine today at Google I.O. The A3 has been purpose-built to handle the considerable demands of these resource-hungry use cases.

Specifically, the company is arming these machines with Nvidia's H-100 GPUs, and combining that with a specialized data center to derive immense computational power with high throughput and low latency. All at what they suggest is a more reasonable price point, then you would typically pay for such a package. If you're looking for specs, consider it's powered by eight Nvidia H-100 GPUs, fourth-gen-intel-Zon-scalable processors, two terabytes of host memory and 3.6 terabytes bisectional bandwidth between the eight GPUs, via NVSwitch and NV-Link 4.0, two Nvidia technologies designed to help maximize throughput between multiple GPUs, like the ones in this product. These machines can provide up to

26 ex-flops of power, which should help improve the time and cost related to training larger machine learning models. What's more of the workloads on these VMs run in Google's specialized Jupiter Data Center networking fabric, which the company describes as 26,000 highly interconnected GPUs. This enables full bandwidth, reconfigurable optical links that can adjust the topology on demand. The company says this approach should also contribute to bringing down the cost of running these workloads. Google will be offering the A3 in a couple of ways. customers can run it themselves, or if they would prefer, as a managed service where Google handles most of the heavy lifting for them, the Do-It Yourself approach involves running the A3 VMs on Google

Kubernetes Engine GKE and Google Compute Engine GCE, while the managed service runs the A3VMs on Vertex AI, the company's managed machine learning platform. While the new A3 VMs are being announced today at Google I.O, they are only available for now by signing up for a preview waitlist, end quote. From the sign of the Times file, some crypto miners are apparently repurposing their GPU rigs to power high-performance computing services for AI clients, quoting Bloomberg. Companies that used and hosted GPUs, or graphics processing units, saw a key part of their once booming business vanish against an increasingly difficult backdrop for crypto. But now mining infrastructure companies like Hive blockchain and Hutt 8 mining are finding opportunities to

repurpose their GPU-based equipment for another industry on the precipice of a possible boom. Artificial intelligence. If you can reapply some of that investment in the GPU mining infrastructure and convert it to new cards and workloads, it makes sense. Hut 8, Chief Executive Officer Jamie Leverton said in an interview, GPUs designed to accelerate graphics rendering require constant maintenance and physical infrastructure not all users are prepared to provide. As such, Hutt 8 and a few other miners have been using the chips to power high-performance computing or HPC services for clients across a range of industries. But inroads with the burgeoning and much-hyped AI sector, which requires huge amounts of computing power, represent the kind

of transformational opportunity miners had been seeking when they originally bought the processors. Hut-8 said its HPC business generated about $16.9 million in 2022, representing about 11% of overall revenue after just one year of operations, driven in part by AI clients. Likewise, Hive blockchain, which purchased $66 million worth of GPUs from Nvidia in early 2021, said it aims to grow its HPC revenue tenfold to $10 million in 2024 and by as much as 20 times current levels by 2025. Currently, analysts tracked by Bloomberg are forecasting about $98 million in overall revenue for the company in 2024. Not all crypto miners are in a position to capitalize on the frenzy around AI or the glut of expensive chips on hand, according to BitPro Consulting, which offers brokerage services for miners between 5 and 15% of

existing crypto-oriented GPUs can be repurposed for AI and adjacent applications like computer vision and generative graphic design. Additionally, a pivot toward offering HPC services for AI will require a huge investment in additional hardware and staff when a lot of miners have been struggling financially with the slide in cryptocurrencies. Core Scientific, the largest public Bitcoin miner by computing power went bankrupt last year, and multiple miners have warned of liquidity crunches. But it could be a chance to recoup the $15 billion that BitPro estimates minors spent on the processors. A handful of ether miners had snapped up the high-end chips when cheaper options were unavailable at the height of the crypto boom in 2021 and early 2022.

The miners were willing to pay more for these overqualified processors because of the favorable economics of soaring ether prices, which touched a high of 4,000. $870 in November 2021. But soon, the group found themselves on the wrong side of a bet regarding not only ether prices, but how long the chips would be useful. Ethereum successfully executed its transition to a proof of stake consensus, making the chips unnecessary in September 2022, not long after Ether had touched a low of $880 following a series of unrelated blowups in the crypto industry, end quote. And finally today, another interesting trend. Tech firms are increasingly hiring in South and Central America because remote hires there are often willing to

accept lower pay than workers in the U.S., but exists in a similar time zone. It has a term called near-shoring, apparently, and quoting from Bloomberg again. So-called nearshoring, hiring or outsourcing to workers in locales closer to a company's home market rather than in distant Asia, is all the rage these days in Mexico where new factories are springing up alongside the U.S. border built by U.S. companies shifting operations from China, but nearshoring is also benefiting countries throughout Latin America that aren't that close to the U.S. Global technology companies are increasingly hiring workers in Chile, Guatemala, Uruguay, and other South and Central American nations to handle such tasks as coding software. Some of these countries can be thousands of miles from the U.S. border,

but they're just a couple of time zones away from much of the U.S. and Europe, making them ideal for handling anything that can be done by phone or computer during the Western Business Day. Global Human Resources Company Deal, which serves clients, including Shopify and Dropbox, estimates that 3,000 U.S.-based companies used its services to hire in Latin America in the first quarter of 2023, twice the number from a year earlier. This is happening while layoffs are dominating headlines in the U.S. as high interest rates increase fears of a recession. About 760 global corporations have slashed more than half a million jobs since October, according to an analysis by Bloomberg in March, with the median layoff, leaving the company

workforce 10% smaller. The technology sector has accounted for about 149,000 jobs lost. At the same time, some companies are adding employees in offshore locations where workers tend to earn less. Deal clients pay an average annual salary of about $74,400 across Latin America for full-time and contract-based workers in areas including engineering and product design. That's compared with a median pay of $127,000 for U.S.-based software developers and $102,000 for a computer systems analyst as of last year, according to the Bureau of Labor Statistics. If for the price of 10 engineers in the U.S., we can hire 100 engineers in Brazil, there's definitely something to think about there, says Alex Boussies, Chief Executive Officer

of Deal. Latin American-based hires are often financially more attractive options. than remote workers in other parts of the world. Deal estimates workers in the region make roughly $20,000 less per year on average compared with Asia-based ones. And while salaries in some popular offshoring locails continue to go up, such as wages in the Philippines soaring 15% year over year in the first quarter, remote workers in Latin America saw wages fall 4% in the same period from a year earlier. Deal says, Latin America produces far fewer engineers per year than an IT outsourcing hotbed such as India. Wormon says that tech employment services,

in Latin America have had to be more proactive about this fact. They're working with local universities and technical schools to learn in advance the profiles of students in the pipeline. Companies looking to hire can also benefit from the recent slowdown in the regional venture capital industry. The region's recent record investments in startups and the subsequent slowdown resulted in a large pool of experienced workers who were laid off from unicorns and are now available for remote work, end quote. Oh man, oh man, oh man, oh man, oh man. Those tears of the kingdom reviews were good. This is going to be a red letter weekend, my friends.

Talk to Tomorrow, which is officially now, Zelda Day.