This Week In Tech
Welcome to the latest edition of our technology newsletter, where we cover some of the most significant events in the tech industry. This week, we delve into the impact of artificial intelligence on the technology landscape, with articles discussing Apple's Siri Chief's struggles as the new AI era begins and the case for big action to regulate AI. We also take a closer look at the effect of ChatGPT, a new AI technology, on 13 software companies and Google's cloud unit's acquisition of a key AI chip team to compete with Microsoft. Additionally, we explore the earnings reports of five big tech companies, including Microsoft's outperformance of Google in Q1 and Discord's revenue growth slowdown. Lastly, we highlight Google's new feature that makes it easier to search for tools in some Google Suite apps and Microsoft's Phone Link app, which now enables iMessage from a PC.
Apple’s Siri Chief Struggles as New AI Era Begins
Three engineers left Apple last year to work on large-language models (LLMs) at Google, which are capable of understanding language and generating humanlike responses. John Giannandrea, a former Google executive Apple hired in 2018, has been accused of organizational dysfunction and a lack of ambition, which have hindered Apple's AI efforts. Siri remains widely derided for its lack of functionality and improvements since Giannandrea took over, and the team building Apple's mixed-reality headset has expressed disappointment in the demonstrations the Siri team created. Apple is the most valuable company in the world, with a $2.6 trillion market capitalization. However, the advent of large-language models (LLMs) such as ChatGPT could fundamentally change how people interact with the online world and create everything from movies to computer programs to slide presentations.
Apple is likely to be more cautious than its competitors in embracing the latest AI technologies, as it is sensitive to maintaining its image and brand. Siri is written, reviewed or edited by humans to avoid embarrassing incidents, and Apple executives have shot down proposals to invest in giving Siri the ability to carry on long back-and-forth conversations. Apple's uncompromising stance on privacy is a challenge, as it has moved many Siri functions to its devices, limiting the number of user queries it transmits to the cloud. Machine learning algorithms are currently in the cloud, but Apple has the hardware expertise and resources to train complex LLMs if it wants to. Siri was introduced in 2011, marking a turning point for AI.
Giannandrea was hired from Google in 2018 to help Apple's Siri group, which had devolved into a mess due to turf battles and heated arguments. He was seen as a major coup for Apple, as he had risen through Google's ranks to become one of its most senior executives, overseeing AI and search. He brought changes to Apple, from his leadership style to policy shifts, and was a contrast to the Type A personalities in Apple's senior ranks.
Giannandrea sought to make Apple a bit more like his old employer, Google, with the goal of recruiting and keeping talent. He created a formal process for Apple researchers to publish papers, increased salaries, and brought in people he worked with at Google. One of his early recruiting successes was the $155 million acquisition of Laserlike, a startup founded by Venkatachary, Baker, and Shukla. The team was among the first to use transformer models, a type of deep learning pioneered by Google. In 2019, Apple released a new Siri feature from the Laserlike team that answered users’ questions by synthesizing information from the web.
The most important details in this text are that Apple acquired other AI startups to bring more expertise in-house, but founders and employees of those startups often left due to Apple's slowness to make decisions or its conservative approach to new AI technologies. Additionally, Giannandrea's new hires have run into seemingly impenetrable bureaucratic obstacles, such as a project to rewrite Siri from the ground up, code-named Blackbird, which competed with the work of two senior Siri leaders, Alex Acero and Robby Walker, who pushed for completion of their own project, code-named Siri X, for the 10th anniversary of the voice assistant. Giannandrea joined Apple to improve Apple's use of data with the goal of making its AI products better. At the time, there were gaps in Apple's collection of Siri metrics due to privacy concerns and lack of internal tools. He approved budgets to expand the number of contractors Apple hired to grade user queries and Siri's responses.
Giannandrea's data science and engineering team grew from a handful of people to more than a hundred, but this arrangement became an issue in mid-2019 when the Guardian reported that Apple contractors were listening to recordings of Siri users without their knowledge. Former employees said the grading effort was a waste of time and money, and the head of the team was asked to find another role.
In early 2019, a 13-year-old boy in Indiana was arrested after posting a screenshot online of Siri recommending a list of nearby schools. Senior Apple leaders demanded an immediate solution and hard-coded new rules into the assistant's software to prevent Siri from responding that way in the future. This is why many former employees in Apple's AI group are skeptical that the company will be successful in developing the next wave of AI products based on LLMs. The Siri design team wanted people to go to Apple's website to see the prices on its products, but engineers clashed over how accurate the responses needed to be before they could release the feature. The design team wanted the feature to be nearly perfect in its answers, while the engineering team pushed for an accuracy rate closer to 80%. In an all-hands meeting, leaders of Apple's AI and machine-learning team agreed that mistakes were bound to happen and that it was their job to educate Apple's top executives about how machine-learning models improve.
How ChatGPT is Roiling 13 Software Companies
Big software companies are racing to incorporate the technology behind ChatGPT, known as generative artificial intelligence, into their products to attract new users and boost profits. However, this technology threatens other companies that have spent years crafting software to automate tasks or build machine-learning models, such as Grammarly, which offers a typing assistant that reviews spelling and grammar. Investors won't fund startups that compete with large-language models unless they have something proprietary to enhance their own versions of the models. Grammarly is tapping generative AI to enhance its offerings, using its own proprietary machine-learning models and data sets to create personalization features in the new product, called GrammarlyGo, that maintain a person's distinct writing voice. Talkdesk, a 12-year-old provider of software that helps companies communicate with customers, released an OpenAI-powered feature in February that automatically summarizes calls between support staff and customers and assesses whether a customer’s issue has been addressed.
It also rephrases customer agent’s chat texts to sound more professional, empathetic or apologetic depending on the circumstances. OpenAI's GPT-3 and GPT-4 language models offer dramatically better performance than the language models Talkdesk has previously used. GitHub has also trained its OpenAI-powered Copilot tool to spot vulnerabilities in code and suggest fixes. The security features launched in February, along with an enterprise version of GitHub Copilot that costs $19 per month per user. Security startups such as Snyk, Qwiet AI (formerly Shift Left) and CircleCI are potentially threatened by generative AI.
Snyk charges at least $57 per month per user, three times Copilot's fee. CircleCI's tools for testing code distinguish it from GitHub Copilot. Stack Overflow has temporarily banned users from posting answers based on GPT or ChatGPT due to a high error rate. Stack Overflow has assigned employees to work on adding generative AI to its products.
Generative AI could pose a threat to startups that automate the process of developing machine-learning models, such as DataRobot, C3 AI, Dataiku and H2O.ai. DataRobot is helping data scientists generate code to train and run machine-learning models, while C3 AI has unveiled an AI-powered service that lets users search for data stored within an organization using natural-language text commands. Databricks has released an open-source large-language model trained on data generated by its employees, while Zapier and Workato have built their businesses around integrating their products with thousands of API integrations from software providers. Alex Clayton, a general partner at Meritech Capital, said generative AI technology could threaten makers of expense-management software such as Expensify by streamlining the process of creating and approving expense reports. Expensify already uses a language model to power its chatbot, Concierge, which responds to questions from 780,000 users at 53,0000 companies.
Daniel Vidal, Expensify's chief strategy officer, plans to tap ChatGPT to improve the chatbot. Rivals are pursuing a similar path, but solutions to the broader challenge of using AI for more complicated tasks are still far off. Kent Bennett believes many companies will find ways to augment their offerings with AI before rivals overtake them.
Discord’s Revenue Growth Slows to 44% in 2022
Discord, a messaging startup, saw revenue growth of 44% to $445 million last year, despite a slowdown from 2021. Revenue grew 26% in the first quarter, but Discord lost $66 million in earnings before interest, taxes, depreciation and amortization. The company has sought to broaden its appeal by removing videogame references and introducing private groups for students. Discord was valued at $14.7 billion in 2021, but recently Fidelity cut its valuation to $9.4 billion. It has been in the news recently due to disclosures that a member of the Massachusetts Air National Guard shared photos of classified documents.
It does not sell ads and is looking for new sources of revenue. News of increased losses could complicate any plans to go public.
Microsoft Outperforms Google in Q
1
Microsoft and Alphabet posted better results than they had for the December quarter. Microsoft reported 7% higher revenue due to growth in its cloud business and commercial software sales. Alphabet had 3% top-line growth due to the stalled digital ad market. Google Cloud's revenue increase of 28% was a decent result, showing the wisdom of Alphabet keeping its investment in the business. Microsoft's success in rolling out ChatGPT-powered advances in its Bing search engine and cloud services positioned it to take market share.
Microsoft Azure matched Google Cloud in revenue growth, and Azure performed better in search with 10% growth. Microsoft stock rose 8.5% in after-hours trading, while Alphabet stock rose just 1.1%. Microsoft's strong quarter and Google Cloud's growth have given reassurance to enterprise software investors, who have been jittery about the extent to which software spending may have decreased.
The Case for Big Action to Regulate Artificial Intelligenc
e
Regulating AI is a necessary and proven boundary condition of capitalism. However, it is premature to regulate something that barely exists, as it creates potential risks such as phishing attacks, hacking access to critical hardware, and the ability to generate politically and economically debilitating fake news stories. It is important to manage and prevent these risks in both the near and long term, as regulation will censor a free and open internet and hinder America’s ability to compete globally. The reality is that the internet is monitored and tracked, not as closely as China’s, but it happens nonetheless. The Children's Online Privacy Protection Act, the General Data Protection Regulation, and the California Consumer Privacy Act are laws that protect internet participants.
America has the opportunity to set and lead global standards instead of allowing them to be written for us. Experience suggests that the FDA employs subject matter experts to evaluate clinical data, manage risk, and take a view on the ultimate benefit to society. Regulating AI is necessary to protect society's best interests and prevent irreparable mistakes. Regulating AI is not to slow down the rate of advancement, but to give ourselves time to evaluate the implications of those advancements. Expectations and guardrails can become a trellis on which later blossomings of life can grow.
AI regulation requires a lasting regulatory framework, smart processes and sound organizational oversight. Gatekeepers already exist in the private sector, such as Apple's App Store and Google Play. A public gatekeeper is needed to investigate new AI models, stress-test them for worst-case scenarios and determine if they are safe for use in the wild. Section 230 was a case study in failing to establish the right regulatory framework for a rapidly evolving industry. The U.S. Supreme Court now has the power to pass reasonable regulation around internet content and its distribution. It is important to consider how to limit AI's potential for harm, as there are mistakes to be made.
Why Tech Rally Left Behind Some Companie
s
The most important details in this text are the reasons why tech stocks have been falling after last year's industrywide battering. These include digital media firm BuzzFeed and electric truck maker Rivian. Bill.com has posted a marked slowdown in top-line growth in the past year, but is still projecting a nearly 50% top-line increase in the March quarter. GitLab has also posted a net loss of $177 million in the first half of its fiscal 2023 year, and its growth in payments handled through GitLab is expected to be flat in the March quarter. Investor jitters about the drastic deceleration are understandable.
GitLab's stock performance has been poor due to its slow growth and low revenue growth rate. The company sells subscriptions to a software platform that makes it easier to develop new software. Clients are taking longer to move ahead on new software purchases and hiring slowdowns and layoffs have reduced the amount of business it has seen from existing customers. GitLab has never generated free cash flow since it went public. Revenue grew faster than operating expenses in the most recent quarter, but profitability still seems like a far-off target.
Warby Parker is a direct-to-consumer eyewear brand, but its aggressive bricks-and-mortar buildout has turned investors off. It is not profitable and hasn't generated free cash flow since the year before it went public. It plans to expand stores and sell more products and services, but profitability will be difficult due to the cost of hiring optometrists. RingCentral's biggest issue may be how competitive a market it plays in. RingCentral is a cloud software-based phone and video communications company that has been on the public markets for a decade. During the pandemic, its stock jumped 120%, but its revenue growth slowed to 25% and this year it has projected revenue will grow at most 11%. Its partners include troubled firms like Avaya, which filed for chapter 11 bankruptcy protection. RingCentral is trying to improve its bottom line by adjusting commission costs for its salesforce to increase profitability margins.
Allbirds is in the wrong direction, with revenue declining by 13% in the fourth quarter of 2022 and 7% growth for the entire year. To turn things around, executives plan to return to more basic products with broader appeal, but it may not be enough to keep Allbirds grounded for good. Farfetch has seen a decline in revenue due to a decline in sales on its e-commerce platform, sales from retailers and brands it owns, and sales from selling tech to other retailers. It has also seen several executive shake-ups recently, including the departure of finance chief Elliot Jordan.
Five Big Tech Companies Report Earnings This Week
This week, five tech companies are expected to report top-line growth averaging just 1.1% for the quarter. This is due to a digital ad market squeeze and weaker corporate spending on cloud services. Wall Street analysts are optimistic that all five will accelerate next year to growth of around 11% to 12% in both 2024 and 2025. Microsoft has lagged its younger counterparts in growth, averaging 14.7% between 2018 and 2022, while Alphabet, Meta and Amazon averaged 23% in the same period. Investors have noticed that Microsoft stock trades at a significant premium to other big tech stocks on a multiple of forward sales.
This week's earnings from Microsoft, Alphabet, Amazon and Meta will provide an update on the state of the digital ad market and online shopping. Microsoft's revenue is expected to be $51 billion +3.4%, while Alphabet's revenue is expected to be $68.8 billion +1.2% and EPS is expected to be $1.07 down from $1.23 a year earlier. Meta Platforms' revenue is expected to be $27.7 billion -0.9%, while Microsoft's revenue is expected to be $51 billion +3.4% and EPS is expected to be $2.24. Meta Platforms' first-quarter report comes as it completes a second wave of mass cuts, with revenue expected to fall as much as 6.8% on the year-ago quarter. Amazon is grappling with a double whammy, with both consumers and companies more cautious about their spending. Snap got burned last year when it had to abruptly lower its second-quarter outlook just a few weeks after issuing it, setting the bar low when it said internal forecasts assumed first-quarter revenue would drop between 2% and 10%.
Snap's forecast is likely to be accurate, with analysts expecting revenue to drop less than 5%, and investors are likely to be listening for any sign of improvement.
Google’s Cloud Unit Gains Key AI Chip Team to Compete With Microsoft
Google has moved the engineering team responsible for making artificial intelligence chips into Google Cloud, a step that could make the cloud unit more competitive with its bigger rivals, Microsoft and Amazon Web Services, in selling AI-powered software to businesses. This is a response to Microsoft and OpenAI's successful launch of ChatGPT, which threatens to erode Google's dominance of search. Google has long developed chips used for AI tasks such as machine learning, known as tensor processing units, which it both uses internally and sells to customers through Google Cloud. The team's move is meant to improve Google Cloud's ability to sell the AI chips to companies that rent its servers and help Google Cloud develop more marketable AI cloud services like the kind Microsoft and OpenAI have begun to sell. Google Cloud has reorganized its team to sell TPUs to cloud customers, bringing all TPU-related work under one leader.
This comes as companies and startups rush to develop AI software that relies on Nvidia chips. Google's TPUs are built for internal purposes with software customized for Google, and Alphabet is adjusting how it allocates costs among its businesses. Google CEO Sundar Pichai hopes Google Cloud can become the company's third profit center.
Google Is Making It Easier To Search For Tools In Some Google Suite App
s
Google is adding a new feature to its Workspace apps that will help users quickly locate commonly used tools without needing to sift through menus and toolbars. The "enhanced tool finder" is appearing as a search bar at the top-left of the app's usual toolbar, featuring a pill-shaped design in keeping with Material Design 3. Tapping on the search bar will re-surface previous queries and keyboard shortcuts will appear beside applicable tools in search results. Google's release timeline indicates it will arrive on Workspace and personal Google accounts by May 25th.
Microsoft's Phone Link App Now Enables iMessage From a PC
Microsoft is rolling out an update to its Phone Link app that includes iOS support for the first time, allowing iPhone owners to connect to a Windows 11 PC and send and receive messages via iMessage, make and receive calls, and see their notifications on a PC or laptop. The update is very basic and only supports making and receiving calls, sending and receiving messages to single contacts, and viewing and dismissing notifications.
Stay informed and up-to-date with the latest technology trends and developments by subscribing to our newsletter.