Tech Brew Ride Home - Fri. 02/03 – The Startup Planning To De-Extinct The Dodo

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, February 3rd, 2023. I'm Brian McCullough today. Earnings wrapups from Alphabet. Investors just want to know when the AI is coming. From Apple. They're still not planning any layoffs. And from Amazon. Look, AWS is still a beast.

Starting point is 00:00:50 And in the weekend long read suggestions, the startup that wants to bring back the dodo, Jurassic Park style. Here's what you miss today in the world of tech. It's that big earnings week for tech. and yesterday was the usual big day with everybody reporting all at once. So let's just rip the Band-Aid off, like we said yesterday, and plow through these. I want to start with Alphabet because there may be the most interesting of the big tech companies right now. Alphabet reported revenue up a mere 1% and net income down a whopping 34%, so decidedly mixed to negative. Overall, Google advertising was down 4% below analyst estimates. Search revenue was down 2%.

Starting point is 00:01:33 even YouTube's revenue dropped 7.8% again below analyst's estimates. The lone bright spot was the all-important cloud revenue segment up 32% to $7.3 billion. So why are investors largely shrugging today? Alphabet's only down a couple of percentage points and the overall NASDAQ is down. So I think the reason that investors are not really worried is because they have their eye on a new ball all sudden how quickly things change. All anyone cares about right now is how Google is going to deal with this AI surge, quoting Market Watch. In a note released Friday, J.P. Morgan analyst Doug Anmuth wrote that AI is reaching an inflection point at alphabet, quote, amidst a challenging operating environment, were encouraged by upcoming AI products and what feels like deeper commitment to

Starting point is 00:02:23 rationalize the cost structure, though there is still some uncertainty around magnitude and timing, he wrote. The analyst explained that AI has been integral to Alphabet's business for many years, driving search and broader ads performance while also supporting its cloud business. After years of heavy investments along with recent competitive pressures, Google will roll out several new AI tools over the next few months, he added, highlighting the company's efforts to layer its Lambda chatbot conversation technology across the company's suite of products, specifically in search and in a possible standalone product. In addition to the company's efforts to deliver AI tools to app developers,

Starting point is 00:02:58 and also pointed to its plans to introduce AI-powered features to workspace and large language models to Gmail and docs. Large language models harness AI to parse vast quantities of natural language, end quote. Indeed, Sundar Pichai said Google will make large language models like Lambda available, quote, in the coming weeks and months, and users will be able to use them, quote, as a companion to search. Quoting Bloomberg, I am excited by the AI-driven leaps we're about to unveil in search and beyond, Pichai said in a statement Thursday, the company has acknowledged the challenges it faces with an increased commitment to efficiency. In January, Alphabet slash 12,000 jobs are 6% of its global workforce, the largest job cuts in its history. We're on an important journey to re-engineer our

Starting point is 00:03:42 cost structure in a durable way and to build financially sustainable, vibrant, growing businesses across Alphabet, Pichai said. Google's ad results align with what its peers are seeing. Executives at SNAP and META, despite reporting quarterly revenue declines, were cautiously optimistic about a return to growth on the horizon for online advertising, end quote. Apple reported Q1 revenue was down 5% year-over-year to $117.2 billion, marking its largest annual quarterly revenue decline since 2016, and net income was also down 13% year-over-year to $30 billion. Tim Cook said iPhone revenue would have grown without supply chain constraints.

Starting point is 00:04:25 Currency fluctuations were a challenge for the company, and layoffs are a, quote, last resort kind of thing. Remember, Apple is basically alone at this point among the big platforms for not having done layoffs thus far. Quoting the journal. The iPhone maker announced its first quarterly revenue decline in nearly four years as manufacturing disruptions in China curbed its ability to deliver premium iPhones. Its results came the same day that Amazon reported growth that while beating expectations, nevertheless concerned investors because of slowdowns in its online shopping and cloud computing businesses and Google Parent Alphabet said it was hit by a broad slowdown in the digital ad market.

Starting point is 00:05:02 So far, Apple has managed to avoid the layoffs rippling through the corporate ranks of its technology peers. The company hired at a slower pace during the pandemic. Over the past three years, Apple's workforce grew about 20% while competitors such as Meta and Amazon almost doubled their employee count in that same period. Mr. Cook said the company is managing costs very tightly and is curtailing hiring in certain areas while continuing to hire in others. I view, layoffs as a last resort kind of thing, Mr. Cook said. You can never say never. We want to manage costs in other ways to the degree that we can, end quote. Mr. Cook said that in addition to manufacturing challenges, the economic climate played a role in the company's results. Quote, we estimated that we

Starting point is 00:05:42 would have grown on the iPhone absent the supply constraints, Mr. Cook said in an interview. For the important holiday quarter, the company struggled to keep up with demand for its latest premium iPhone 14 pro models, as China's zero COVID policies caused upheaval. at a smartphone factory in Zhang Zhao, end quote. And real quick, let's hit Amazon's numbers. They reported Q4 revenue up 9% year-over-year. Net income was down 98% to a mere $278 million. It was their least profitable holiday quarter since 2014.

Starting point is 00:06:19 AWS revenue was up 20% year-over-year to $21.4 billion for the quarter. Q4 advertising services revenue grew 19% and subscription services grew 13%. spending on content actually surged to $16.6 billion last year, up 28% largely because of their big Thursday night football deal. But again, look, AWS continues to be the thing here. Google Cloud is at $7.3 billion in quarterly revenue, while AWS, as I just said, is triple that at $21.4 billion a quarter. Amazon's AWS revenue was $80 billion in 2022, higher than the revenue of 457,000 of the companies in the S&P 500. They've gone from doing $3 billion a year in revenue

Starting point is 00:07:05 to $80 billion a year in revenue in just nine years, a 44% annualized growth rate. This headline just came out after I recorded the Longreed segment, so if you hear me mention Anthropic in that segment without sounding like I knew this was coming, it's because I didn't. The Financial Times has sources that are saying Google has invested $300 million in Anthropic,

Starting point is 00:07:31 taking a 10% stake in that generative AI startup, and requiring it to spend money on Google Cloud. Quoting the FT. The terms of the deal through which Google will take a stake of around 10% requires Anthropic to use the money to buy computing resources from the search company's cloud computing division, according to three people familiar with the arrangement. Google's move highlights the influence that a small number of big tech companies

Starting point is 00:07:53 have assumed over other companies working on AI, which need access to cloud computing platforms, to handle the giant AI models developed by groups such as Anthropic. Anthropic was formed in 2021 when a group of researchers led by Dario Amodi left OpenAI after a disagreement over the company's direction. They were concerned that Microsoft's first investment in OpenAI would set it on a more commercial path and detract from its original focus on the safety of advanced AI. Anthropic has developed an intelligent chatbot called Claude, rivaling OpenAI's chat GPT, though it has not yet

Starting point is 00:08:26 been released publicly. The startup had already raised more than $700 million before Google's investment, which was made in late 2022, but is not previously been reported. The company's biggest investor is Alameda Research, the crypto hedge fund of FTCS founder Sam Bankman-Fried, which put in $500 million before filing for bankruptcy last year. FTX's bankruptcy estate has flagged Anthropic as an asset that may help creditors with recoveries, end quote. Time for the week on long read suggestions. Look, what do you think I have for you this week? I'm hustling as much as everybody to stay on top of these AI developments, so here you go, a couple of primers that will help, you know, keep us abreast of the lay of the land.

Starting point is 00:09:14 From Ars Technica, a look at the history of generative AI and developments that paved the way for the most recent breakthroughs, including Kuta, convolutional neural networks and transformers. Basically, how did we get here? Why is it happening now? Quote, there's a reason all of this has come at once. The breakthroughs are all underpinned by a new class of AI models that are more flexible and powerful than anything that has come before, because they were first used for language tasks like answering questions and writing essays, they're often known as large language models, OpenAI's GPT3, Google's BERT, and so on are all LLMs. But these models are extremely flexible and adaptable, the same mathematical structures that have been so useful in computer

Starting point is 00:09:53 vision, biology, and more that some researchers have taken to calling them foundational models to better articulate their role in modern AI. There's a holy trinity in machine learning, models, data, and compute. Models are algorithms that take inputs and produce outputs. Data refers to the examples the algorithms are trained on. To learn something, there must be enough data with enough richness that the algorithms can produce useful output. Models must be flexible enough to capture the complexity in the data, and finally, there has to be enough computing power to run the algorithms. The first modern AI revolution took place in deep learning in 2012 when solving computer vision problems with convolutional neural networks or

Starting point is 00:10:33 CNNs took off. CNNs are similar in structure to the brain's visual cortex. They've around since the 1990s, but weren't yet practical due to their intense computing power requirements. In 2006, though, Nvidia released Kuda, a programming language that allowed for the use of GPUs as general-purpose supercomputers. In 2009, Stanford AI researchers introduced ImageNet, a collection of labeled images to train computer vision algorithms. In 2012, AlexNet combined CNNs trained on GPUs with ImageNet to create the best visual classifier the world had ever seen. deep learning and AI exploded from there, end quote. And from Venture Beat, a roundup of potential rivals to open AI's chat GPT service,

Starting point is 00:11:16 DeepMind's Sparrow, Google's Lambda, Anthropics Claude, and Character AI, quote, Anthropic developed an AI chatbot Claude, available in closed beta through a Slack integration, that reports say is similar to chat GPT and has even demonstrated improvements. Anthropic, which describes itself as, quote, working to be able to. build reliable, interpretable, and steerable AI systems, created Claude using a process called constitutional AI, which it says is based on concepts such as beneficence, non-malfeasance, and autonomy. DeepMind, which is the British-owned subsidiary of Google Parent Company Alphabet, introduced Sparrow in a paper in September. It was hailed as an important step toward creating

Starting point is 00:11:57 safer, less biased machine learning systems, thanks to its application of reinforcement learning based on input from human research participants for training. DeepMind says Sparrow is a, quote, dialogue agent that's useful and reduces the risk of unsafe and inappropriate answers, end quote. The agent is designed to, quote, talk with a user, answer questions and search the internet using Google when it's helpful to look up evidence to inform its responses, end quote. Google's Lambda is still considered to be one of ChatGPT's biggest competitors. Launched in 2021, Google said in a launch blog post that Lambda's conversational skills have been years in the making. Like ChatGPT, Lambda is built on Transformer, the neural network architecture that Google

Starting point is 00:12:36 research invented and open-sourced in 2017. The Transformer architecture, quote, produces a model that can be trained to read many words, a sentence or paragraph, for example, pay attention to how those words relate to one another, and then predict what words it thinks will come next, end quote. And like ChatGPT, Lambda was trained on dialogue, according to Google. During its training, Landa picked up on several of the nuances that distinguish open-ended conversation from other forms of language, end quote. Finally, what happens when engineers who developed Google's Lambda get sick of big tech bureaucracy and decide to strike out on their own? Well, just three months ago, Nome Shazir, who was also one of the authors of the original Transformer paper, and Daniel DeFritus

Starting point is 00:13:16 launched Character AI, its new AI chatbot technology that allows users to chat and roleplay with, well, anyone, living or dead. The tool can impersonate historical figures like Queen Elizabeth and William Shakespeare, for example, or fictional characters like Draco Malfoy. According to the information, Character has told investors that it wants to raise as much as $250 million in new funding, a striking price for a startup with a product still in beta. Currently, the report said the technology is free to use and character is, quote, studying how users interact with it before committing to a specific plan to generate revenue, end quote. The Atlantic has a mind-bending piece about how the human brain could explain why AI programs are so good at writing grammatically superb nonsense. Quote, chat GPT and software like it demonstrate an incredible ability to string words together,

Starting point is 00:14:05 but they struggle with other tasks. Ask for a letter explaining to a child that Santa Claus is fake, and it produces a moving message signed by St. Nick himself. These large language models, called LLMs, work by predicting the next word in a sentence based on everything before it, but ask chat GPT to do basic arithmetic or spelling or give advice for frying an egg, and you may receive grammatically superb nonsense. Quote, if you use too much force when flipping the egg, the egg shell can crack and break. quote. These shortcomings point to a distinction not dissimilar to one that exists in the human brain, between piecing together words and piecing together ideas, what the author's term formal and

Starting point is 00:14:40 functional linguistic competence, respectively. Language models are really good at producing fluent grammatical language, says the University of Texas at Austin linguist Kyle Malwold, the paper's other lead author, but that doesn't necessarily mean something which can produce grammatical language is able to do math or logical reasoning or think or navigate social contexts. And quote. If the human brain's language network is not responsible for math, music, or programming, that is, for thinking, then there's no reason an artificial neural network trained on terabytes of text would be good at those things either. In line with evidence from cognitive neuroscience, the authors write, LLM's behavior highlights the difference between being good at language and being

Starting point is 00:15:18 good at thought. ChatGPT's ability to get mediocre scores on some business and law school exams, then, is more a mirage than a sign of understanding, end quote. We didn't get to cover this this week, but you might have heard about the big raise recently for that startup that wants to bring back the dodo. Yes, the extinct bird. Colossil biosciences of Austin, Texas is two years old and just raised a huge $150 million round, bringing its total raise to $225 million. Yes, this is Jurassic Park stuff. Quoting MIT Technology Review. The resurrection of the dodo is a theoretical possibility thanks to Beth Shapiro, a specialist in ancient DNA at the University of California Santa Cruz, who says that,

Starting point is 00:16:00 that she and coworkers were able to recover detailed DNA information from a 500-year-old dodo remains held at a museum in Denmark. I have the dodo genome, Shapiro, who is now advising colossal said in a phone interview with MIT Technology Review. That is something we just finished, end quote. To create the dodo from such genetic information, the company plans to try to modify the bird's closest living relative, the brightly colored Nicobar Pigeon, turning it step by step into a dodo and possibly rewilding the animal in its native habitat. Colossal has not yet created any kind of animal, though. It's still working on developing the necessary processes, and making a dodo might not even be possible. That's because it's hard to predict how many DNA changes will be needed to

Starting point is 00:16:42 transform the Nicobar Pigeon into a big-beaked three-foot-tall dodo, end quote. And finally, final link in the show notes, just for fun, a look at the classic calculators that now are emulated on the Internet Archive. Yes, like the Internet Arcade, there's a whole emulator movement to preserve those Hewlett Packers and Texas Instruments, baddies that some of us are old enough to actually remember using. No bonus episode this weekend, or probably next weekend too, but I do have an ask, and I know I've had a ton of asks for you lately.

Starting point is 00:17:23 Come to the comedy show, let me know the cool AI tools you're seeing, but this one is a bit more nuanced, but also a bit more urgent. You all have heard my struggles to fill ads on this show lately. It's largely because the ad network we've relied on for years has suddenly stopped performing. So I've sort of been looking at switching networks, but given the complexity of doing that, I've also thought, what if I just built what I want myself, what I need for myself? So I'm looking to hire a salesperson to sell ads on this show. We'd start small, a monthly stipend, variable compensation in terms of percentage of ad money closed.

Starting point is 00:17:58 Basically, whomever took this job could start sort of as a side gig. If we were successful, it could very much grow into a full-time job if you sold out the inventory just for this show in a month. By my math, the job would pay around 9,000 or so a month in terms of the variable compensation. Obviously, we might not get to there, but I want to see how far we can get. And also, I'm not the only one who's having trouble with ads lately. there are other podcasters and sub-stackers in the tech space who could use some ads sold as well. We could scale this thing up into a full-fledged sort of operation where you could sell for a bunch of us all doing the tech stuff. Obviously, ad sales experience would be a requirement.

Starting point is 00:18:43 Experience selling podcast ads would be a definite plus, but experience selling to tech companies. You know, the sort of companies that advertise on here when times are good, the companies that want to reach other startup companies and tech employees. Experience with that sort of thing would be a double plus. I could help guide you in this area because I have an idea of the companies we should be trying to sell to. I just don't have the time to do it myself. Anyway, if this sounds interesting to you or if you know someone for whom it might be interesting, hit me up at brian at techmeme.com and let's discuss. Final ask for a while, I hope.

Starting point is 00:19:19 Talk to you on Monday.

Tech Brew Ride Home - Fri. 02/03 – The Startup Planning To De-Extinct The Dodo

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