Gone: DVDs, Magazines, Buzzfeed News
Gone: DVDs, Magazines, Buzzfeed News
That Was The Week 2023, #14
A reminder for new readers. That Was The Week collects the best writing on critical issues in tech, startups, and venture capital. I select the articles because they are of interest. The selections often include things I disagree with. The articles are only snippets. Click on the headline to go to the original. I express my point of view in the editorial and the weekly video.
My Friend Yuri Rabinowitch is hosting an Investors’ roundtable next week in Silicon Valley. Click below to attend.
Content this week from: @jonfingas, @harrymccracken, @TurnerNovak, @fredwilson, @kwharrison13, @ballmatthew, @ttunguz, @timoreilly, @davidcummings, @NeilThanedar
Contents
Editorial: Gone!
Essays of the Week
AI of the Week
News Of the Week
Elon Musk wants to develop TruthGPT, ‘a maximum truth-seeking AI’
Google Looks to Turbocharge AI Efforts With Combined Brain, DeepMind Unit
Tiger Global Management’s $12.7 Billion Venture Fund Records 20% Loss
Apple launches Apple Card’s savings accounts with 4.15% interest rate
Startup of the Week
SeetGeek
Tweet of the Week
@NeilThanedar
Editorial
This week’s headline is “Gone.” Specifically, Nexflix’s DVD service, Maximum PC and MacLife Magazines, and BuzzFeed News. The latter was “shuttered” today.
We could extend the meaning to include unicorns, later-stage venture funding, and Elon Musk’s first Starship.
In the case of the first three, it is surprising that they survived this long.
But I think these closures are a moment to acknowledge and mark. They have already been replaced long ago, and their passing is a testament to the work Netflix has done with streaming, the emergence of the newsletter and blog, and the frustrations we all have with headline writers seeking clickbait.
The downbeat tone continues with a Vox piece asking (with an implied answer) Is Twitter finally dying? Vox is no longer a news publication when it comes to Twitter, it has become a rag with a campaign seeking to fight against the company’s very existence. It is the Fox News of Twitter coverage.
A more thoughtful piece from Kyle Harrison is still, however, downbeat. It asks why almost all unicorns have never made a profit but overlook that many have enormous free cash flow that they decided to spend on growth as a conscious strategy. Spending profits (thus removing them as profits) is what startups should be doing. Amazon has done so for decades successfully.
Turner Novak writes about Substack, one of the emergent platforms, and asks the question, “Can Substack Save the Social Network?”
He tracks the history of Substack, increasing the tools that enable its writers, creators, and consumers to discover and recommend each other and the impact on subscriber growth. Meanwhile, Tomasz Tunguz of Theory VC states that all startups must have an AI strategy. Fred Wilson says protocols are important but are not well understood.
These articles and others focus on what is new and what comes next. As Leonard Cohen famously wrote, “that’s how the light gets in.”
In Silicon Valley, over $11.5 billion was invested in AI startups in the past three months. Web 3 funding has slowed to a crawl but has not disappeared. And biotech funding remains vibrant, along with new energy startups.
We are entering a phase where a new cohort of companies, funded since the middle of 2022, is attracting capital because of the scale of their ambition and the opportunity it represents. The number of funding rounds at the seed and early stage is declining, but the quality of the investments and the amounts raised are both strong.
2023 will be a big year for the next generation of successful startups and I can’t wait to see what they bring to the table.
Essays of the Week
Netflix will shut down its DVD rental business in September
The company's original business is coming to an end.
![Reed Hastings, CEO of Netflix Inc., the online DVD-rental service, holds several DVDs as he poses at the Netflix offices in Beverly Hills, California, December 8, 2005. [Netflix has come out on top in a yearlong struggle against rival Blockbuster Inc. ]](https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffb1a6a7c-dd3c-47eb-aa39-fa3ada6f18d0_875x612.jpeg "Reed Hastings, CEO of Netflix Inc., the online DVD-rental service, holds several DVDs as he poses at the Netflix offices in Beverly Hills, California, December 8, 2005. [Netflix has come out on top in a yearlong struggle against rival Blockbuster Inc. ]")
Fred Prouser / reuters
Jon Fingas| @jonfingas|April 18, 2023 3:52 PM
After 25 years, Netflix's original business is shutting down. The company has revealed that it will "wind down" DVD rentals (that is. DVD.com), with its last movie discs mailing on September 29th. Simply put, the shrinking demand for physical rentals is making it "increasingly difficult" to offer the quality of service the company wants.
Netflix shipped its first disc (Beetlejuice, if you're curious) in 1998. It has since mailed over 5.2 billion movies in its signature envelopes (nearly all of them before 2019) to more than 40 million customers. You likely know the story after that. The company began streaming on-demand video in 2007, and that business grew quickly enough that it became Netflix's most popular offering by 2009. After a premature attempt to spin off the mailed rentals as Qwikster in 2011, Netflix moved them to DVD.com in 2016. By that point, the company was well into producing original streaming shows and had stopped mailing DVDs on Saturdays.
There's also a financial incentive to drop disc rentals. While Netflix is recovering from a bleak 2022, its profits in the first quarter were still worse than they were a year ago. Subscriber growth was relatively modest, too, at 1.7 million new users. A decision to axe the DVD-by-mail unit could help Netflix trim costs, even if the savings are relatively small. The disc market has been on a sharp and steady decline for years, according to VideoScan/MediaPlayNews data — sales alone dropped 19 percent from 2021 to 2022.
The End of Computer Magazines in America
With Maximum PC and MacLife’s abandonment of print, the dead-tree era of computer journalism is officially over. It lasted almost half a century—and was quite a run.
Posted by Harry McCracken on April 15, 2023 at 9:22 am
The April issues of Maximum PC and MacLife are currently on sale at a newsstand near you—assuming there is a newsstand near you. They’re the last print issues of these two venerable computer magazines, both of which date to 1996 (and were originally known, respectively, as Boot and MacAddict). Starting with their next editions, both publications will be available in digital form only.
But I’m not writing this article because the dead-tree versions of Maximum PC and MacLife are no more. I’m writing it because they were the last two extant U.S. computer magazines that had managed to cling to life until now. With their abandonment of print, the computer magazine era has officially ended.
The first issue of Byte, the first magazine about personal computers—and many people’s candidate for the best such publication, period..
It is possible to quibble with this assertion. 2600: The Hacker Quarterly has been around since 1984 and can accurately be described as a computer magazine, but the digest-sized publication has the production values of a fanzine and the content bears little resemblance to the slick, consumery computer mags of the past. Linux Magazine (originally the U.S. edition of a German publication) and its more technical sibling publication Admin also survive. Then again, if you want to quibble, Maximum PC and MacLife may barely have counted as U.S. magazines at the end; their editorial operations migrated from the Bay Area to the UK at some point in recent years when I wasn’t paying attention. (Both were owned by Future, a large British publishing firm.)
Still, I’m declaring the demise of these two dead-tree publications as the end of computer magazines in this country. Back when I was the editor-in-chief of IDG’s PC World, a position I left in 2008, we considered Maximum PC to be a significant competitor, especially on the newsstand. Our sister publication Macworld certainly kept an eye on MacLife. Even after I moved on to other types of tech journalism, I occasionally checked in on our erstwhile rivals, marveling that they somehow still existed after so many other computer magazines had gone away.
Shuttering of BuzzFeed News Signals Shift to Survival Mode
By
Martin Peers, Sahil Patel and Paris Martineau
April 20, 2023 11:24 AM PDT ·
BuzzFeed, like much of the digital media industry, is moving into survival mode.
BuzzFeed said Thursday it would shutter BuzzFeed News, the highest-profile part of its business but one that loses money. The move is part of a broader effort to reduce losses, which also includes laying off 15% of its staff, or 180 people. The cutbacks spotlight the financial pressures on BuzzFeed and are likely to intensify investor pressure on CEO Jonah Peretti to sell the company.
BuzzFeed, which primarily makes money from videos it distributes on social media such as through its Tasty cooking channel, now has an enterprise value of about $200 million.
THE TAKEAWAY
• Small digital media firms suffering 20-30% drops in ad in Q1
• BuzzFeed has been burning cash
• Investors want Peretti to sell BuzzFeed
Peretti, who founded BuzzFeed in 2006 and has a controlling stake, is showing no signs of looking to sell, however. In a memo to staff, Peretti said he plans to “fight our way to a bright future.” A person close to the company said BuzzFeed’s divisions, aside from BuzzFeed News, are profitable.
The closure of BuzzFeed News reflects a broader retrenchment across the digital media industry as firms reckon with a severely depressed digital ad market. Also Thursday, Axel Springer’s Insider said it would lay off 10% of its staff, the latest in a series of staff cuts the German media company instigated at its U.S. digital media holdings. The cuts included closure of its Protocol tech news site. Meanwhile, Vice Media, another once highly valued new-media firm, has been on the market amid a rash of top executive exits.
While the ad slowdown has buffeted digital ad giants such as Meta Platforms and Alphabet’s Google, it has hit smaller media firms much harder. BuzzFeed reported a 27% drop in advertising for the fourth quarter of 2022. Conditions have not improved. In the first quarter of this year, some smaller digital media firms have experienced ad revenue declines of 20% to 30%, according to executives at three digital media firms.
As Insider founder Henry Blodget put it to employees at an all-hands staff meeting today, the industry is dealing with a “significant recession in advertising spending,” according to a recording of the meeting reviewed by The Information.
Insider, like BuzzFeed, relies on big tech companies like Meta Platforms’ Facebook and Google to distribute its content. And changes those companies have made have hurt the smaller digital media firms. Facebook, for instance, has moved away from longer videos made by firms like BuzzFeed in favor of the shorter Reels format, which does not offer publishers a way to make much advertising revenue. These kinds of shifts are “affecting both our distribution and the revenue share that we get from a lot of the platforms,” said Blodget during the meeting.
Can Substack Save the Social Network?
Plus new products from Stripe, AngelList, and Epic Games
APR 11, 2023
This past week, Substack dropped what may have felt like an insignificant new feature, Notes. But it lined up with something I’ve been thinking about for awhile: is Substack the next big social network?
I’ve had this draft half-written for a few months. It’s admittedly a bit crazy to even propose, and it’s definitely not how Substack describes itself. But a series of features and deliberate product choices over the years (and the fact Twitter recently started aggressively muting any mention of Substack in its feed) lead me to think it has a chance.
Substack: The Subscription Social Network
Society if everyone paid for at least one Substack subscription
Substack just announced a new feature, Notes. It seems small, as it’s essentially “tweets for Substack”. But it scared Twitter enough to suppress any mentions of Substack in outbound links. In my opinion, this continues Substack’s march towards becoming what most people would call a social network - and possibly something much bigger.
If you’re not familiar with Substack, it started in 2017 as the best tool for writing a blog. It had a simple editor that was easy to use, generated great SEO ranking out of the box, and placed heavy emphasis on email collection and the reader experience.
I’m probably getting the exact order of operations wrong, but Substack then enabled writers to paywall specific content (or maybe it started with this?), giving the tools to run a premium content business. Among many other features, it added things like podcasts and video support, cross-newsletter recommendations, and eventually a mobile app (and chat).
If you’ve followed the Creator Economy over the past three years, it became a graveyard of startups that never quite solved an urgent enough problem. Looking across the landscape of creator-focused products, the first tier (at least in terms of valuation) include YouTube, Instagram, and TikTok. These are each worth north of $100 billion, primarily because they help creators solve their most urgent problem: distribution. We can argue how well they actually solve this problem, but it does make these platforms almost impossible for creators to leave.
Next is the second tier of creator-related businesses, which typically help solve for monetization. This means they have a clear path to generating revenue, and includes companies like Patreon and Kajabi. While they enable creators to run a business, they don’t generally help grow them grow them, which is the most urgent problem for most content creators. If you only read a sentence or two about what Substack does, you’d probably think a tool for publishing a blog ends up in this second tier. But that would be incorrect.
Substack started helping creators with distribution a few years ago. In its early days it had a public leaderboard featuring its top writers, “recommended for you” articles below each post, a feed featuring all of a reader’s subscriptions, and eventually a standalone mobile app. Substack supercharged its distribution when it launched recommendations in 2022. You can tell exactly when this happened based on Lenny’s newsletter growth below.
Lenny is a former startup founder and spent seven years in growth and product roles at Airbnb. Or in other words, the exact person you’d expect to be good at growing a newsletter. He was already growing steadily, and recommendations added rocket fuel.
The boost from recommendations is even more noticeable looking at Newcomer, run by former Bloomberg journalist Eric Newcomer. It’s probably not a stretch to say Substack recommendations changed his life…
APR 17, 2023
What Is A Protocol And Why Does It Matter?
A VC by Fred Wilson
Enabling trusted brands that broaden access to knowledge, capital, and well-being by leveraging networks, platforms, and protocols.
https://www.usv.com/#thesis-3-0
That last word is powerful but unfortunately less understood than the other words in that sentence.
Protocols have been around forever and are well-understood codes of conduct between people.
In computer science, protocols are the same thing, codes of conduct.
I like this definition of computer protocols from the Cloudflare website:
Standardized protocols are like a common language that computers can use, similar to how two people from different parts of the world may not understand each other’s native languages, but they can communicate using a shared third language. If one computer uses the Internet Protocol (IP) and a second computer does as well, they will be able to communicate — just as the United Nations relies on its 6 official languages to communicate amongst representatives from all over the globe. But if one computer uses IP and the other does not know this protocol, they will be unable to communicate.
https://www.cloudflare.com/learning/network-layer/what-is-a-protocol/
Protocols have been around for as long as computers, but we are at the beginning of a golden era of protocols that I like to call “web3.”
Until recently protocols were mostly free and unmonetized. We all use the HTTP protocol every day to access web pages. We all use the SMTP protocol every day to send email. But these protocols are free to use and don’t make money for any company or project or individual.
With the arrival of Bitcoin back in January 2009, we got a protocol that had monetization built in.
And since then, computer scientists have been creating web3 protocols for all sorts of things, most of which have a token that monetizes the protocol.
Building an Actual Unicorn
$1B Valuation < $1B ARR < $1B Cash Flow
APR 15, 2023
A Tweet Heard Round My Head
It's always the sign of a great tweet when it leaves you thinking about it so much that you put off the thing you were planning on writing about and write about something totally different. Even when your baby was up until 1 AM and it's Saturday and you haven't started writing anything. You still can't bring yourself to write about anything else. That's how I felt this week after a tweet from Matthew Ball.
If you don't know who Matthew is, he's absolutely worth a follow. He's worked in media and entertainment, is one of my favorite writers on gaming, and wrote a book about the metaverse.
So what's the mind-consuming tweet in question?
The replies are fascinating. Not only because people often just whip stuff out on Twitter without thinking about it, but also because of the broad and diverse set of perspectives. By asking one question you expand across industries, into discussions of stock-based compensation, and the value of different measures of profitability.
I've seen similar discussions about different types of companies that get built (side note: if you're not following Aashay, Ho Nam, and Post Market, that's a real loss. Here's a peek.)
Now, I'm no public analyst. There are a lot more thoughtful in-the-weeds analysts out there than me, many of them show up in the comments on those threads. So my head wasn't just focused on answering the question. The crowd-sourced efforts from the comments has led to this follow-up list from Matthew.
More than that, it left me reflecting on the long-run outcome of the current venture ideology. So I wanted to spend a second trying to articulate my reactions, justifications, disappointments, and resolutions……
2023 Private SaaS Company Valuations
April 14, 2023
Long-time readers of our work know we have strived to shed an objective light on the opaque, confusing, and volatile practice of valuing private SaaS companies. As a provider of debt financing for private B2B SaaS companies since 2007, SaaS Capital has provided funding to more than 100 firms. This puts us in a unique position to witness equity raises and M&A events. Using this information, we published our first private SaaS valuation framework in 2016 and, as the market has evolved, we have iterated on the methodology.
In Q2 2022, SaaS Capital released the most recent update on how to value private SaaS companies. This data-driven methodology is based on a statistical analysis of over ten years of data. While there are dozens of other factors impacting valuation, the research resulted in a downloadable Excel model using three primary variables that drive valuations:
The current public market valuations.
Annual recurring revenue (ARR) growth rate.
Net revenue retention.
The current white paper can be found here – What’s Your SaaS Company Worth?
2023 Public SaaS Valuation Multiples
Public market valuations reflect real-time information and have high data integrity because they include many different companies and are based on audited financial statements. Public valuation data is the primary starting point for valuation analysis by both buyers and sellers.
We developed the SaaS Capital Index™ (SCI) to represent what we feel is the best profile of a B2B SaaS company. The revenue multiple is based on annualized current run-rate revenue, not trailing or projected revenue. We believe run-rate revenue is the most accurate and objective measure of the current scale of the business and, therefore, the best measure to be used for valuation purposes. The index excludes SaaS companies serving B2C customers and very small B2B companies with annual revenue per customer of less than $500. Companies targeting these end users have customer acquisition and retention dynamics that are significantly different than those of traditional B2B SaaS businesses. The SCI is updated monthly and free to download here. As of March 31, the SCI shows a median public valuation multiple of 7.1.
ARR Growth Rate
ARR growth rate is a crucial company-specific metric, and it’s important to calculate it correctly, not merely to use a generalized estimate of “revenue.” ARR Growth Rate should be annualized, actual, and trailing (not “forecast” or “projected”). ARR Growth Rate is highly correlated with valuation multiples, although we can still make a better estimate by using it in combination with other factors.
Is Twitter finally dying?
Say goodbye to the old Twitter and hello to the new normal.
By Shirin Ghaffary Apr 15, 2023, 6:00am EDT
It’s been a year since Elon Musk initiated his takeover of Twitter. The six months since he actually took charge can only be characterized as chaotic, and quietly, in early April, Musk merged Twitter with a new shell company called X Corp. In other words, Twitter Inc. no longer exists.
Musk painted a rosy picture in an April 12 interview with the BBC. He said that Twitter is thriving with “record high usage” and that, despite some “compute glitches here and there,” the “site is doing really well,” advertisers who initially fled the platform after his takeover have mostly returned, and the company is on track to make a profit by next quarter. Musk also scoffed at the idea that he had destroyed Twitter, saying that predictions that the platform would “cease to exist” have “turned out to be false.”
To some extent, what Musk was saying is true: Twitter did not implode overnight in a catastrophic technical meltdown, as some of the more skeptical critics of Musk’s reign had warned.
So Twitter isn’t dead. But it’s getting there.
If you’re on Twitter today, you’re likely experiencing an app that is more frequently broken, more random, and more unhinged. The site has seen a sustained uptick in outages and bugs. Many users are complaining about seeing more content in the new For You feed that they don’t want to see. There are active neo-Nazis on the platform. And just this week, NPR and PBS, two major news outlets with over 10 million combined followers, stopped using the platform, saying they had lost faith in Musk’s decision-making after he slapped a “government-funded media” label on the accounts.
Some people on Twitter, especially Musk’s dedicated fan base, are embracing Musk’s shake-up. But there’s no denying that Twitter today is a more unreliable experience than it was before Musk took over. Musk has justified his many controversial moves by saying that he’s doing it in the name of free speech: to remove power from check mark-touting “media elites” and distribute it to everyone else, no matter how abhorrent their views.
The problem is, Musk’s plan doesn’t seem to be working as well as he’s saying. Twitter’s web traffic dropped by nearly 8 percent last month compared to the year before, and has been dropping for the past three months year over year, according to new estimates from data intelligence firm Similarweb. This directly contradicts Musk’s claims that usage is up. In terms of Twitter’s main line of income, advertising, outside data also conflicts with Musk’s claims that all is well. Multiple reports over the past few months have shown that top brands are continuing to leave the platform since Musk took over…..
AI of the Week
Why Every Startup Needs an AI Strategy
Venture Capitalist at Theory
Every company will be an AI company. Every startup needs an AI strategy - not just for fundraising or press appeal.
User expectations have changed. When one email composer window autocompletes sentences, every email product will need to follow. When one customer support bot provides a meaningfully better experience to answer questions, every competitor will match it.
ChatGPT & Midjourney have educated hundreds of millions that we can & should expect more from software. Now those teeming masses will demand it.
How quickly users switch from one product to another depends on the category of software : the switching costs, the pain associated with a lack of the feature, or the loss in productivity without it.
Today, there are three different levels of AI strategy:
single feature : autocompletion of text or code within a text editor
department level : automated customer support agent
core to the product : an image generation business that builds a proprietary large language model
Most businesses will likely adopt the first & as the technologies mature, the second. Newer businesses will pursue the third as a competitive advantage as they reimagine workflows.
Many core software products map real world workflows to software ; these will change.
Text & chat UIs will change those expectations : imagine developing a project management plan in a conversation with a computer or reconciling expenses by dictating an answer to a mobile phone.
These new user interfaces will change software.
It may be easy to dismiss an AI strategy as a fundraising tactic (yet another AI company), but the trope underscores a meaningful shift in what ultimately matters :
User expectations have evolved & every business will need to meet their demands to thrive.
Sam Altman: Size of LLMs won’t matter as much moving forward
Ron Miller @ron_miller / 8:57 AM PDT•April 14, 2023
Image Credits: Steve Jennings / Getty Images
When OpenAI co-founder and CEO Sam Altman speaks these days, it makes sense to listen. His latest venture has been on everyone’s lips since the release of GPT-4 and ChatGPT, one of the most sophisticated large language model-based interfaces created to date. But Altman takes a deliberate and humble approach, and doesn’t necessarily believe that when it comes to large language models (LLM), that bigger is always going to be better.
Altman, who was interviewed over Zoom at the Imagination in Action event at MIT yesterday, believes we are approaching the limits of LLM size for size’s sake. “I think we’re at the end of the era where it’s gonna be these giant models, and we’ll make them better in other ways,” Altman said.
He sees size as a false measurement of model quality and compares it to the chip speed races we used to see. “I think there’s been way too much focus on parameter count, maybe parameter count will trend up for sure. But this reminds me a lot of the gigahertz race in chips in the 1990s and 2000s, where everybody was trying to point to a big number,” Altman said.
As he points out, today we have much more powerful chips running our iPhones, yet we have no idea for the most part how fast they are, only that they do the job well. “I think it’s important that what we keep the focus on is rapidly increasing capability. And if there’s some reason that parameter count should decrease over time, or we should have multiple models working together, each of which are smaller, we would do that. What we want to deliver to the world is the most capable and useful and safe models. We are not here to jerk ourselves off about parameter count,” he said.
Altman has been such a successful technologist partly because he makes big bets, and then moves deliberately and thinks deeply about his companies and the products they produce — and OpenAI is no different.
“We’ve been working on it for so long, but it’s with gradually increasing confidence that it’s really going to work. We’ve been [building] the company for seven years. These things take a long, long time. I would say by and large in terms of why it worked when others haven’t: It’s just because we’ve been on the grind sweating every detail for a long time. And most people aren’t willing to do that,” he said.
When asked about the letter that requested that OpenAI pause for six months, he defended his company’s approach, while agreeing with some parts of the letter.
You Can’t Regulate What You Don’t Understand
Or, Why AI Regulations Should Begin with Mandated Disclosures
By Tim O’Reilly
April 14, 2023
Robot love (source: Pixabay)
Learn faster. Dig deeper. See farther.
Join the O'Reilly online learning platform. Get a free trial today and find answers on the fly, or master something new and useful.
The world changed on November 30, 2022 as surely as it did on August 12, 1908 when the first Model T left the Ford assembly line. That was the date when OpenAI released ChatGPT, the day that AI emerged from research labs into an unsuspecting world. Within two months, ChatGPT had over a hundred million users—faster adoption than any technology in history.
The hand wringing soon began. Most notably, The Future of Life Institute published an open letter calling for an immediate pause in advanced AI research, asking: “Should we let machines flood our information channels with propaganda and untruth? Should we automate away all the jobs, including the fulfilling ones? Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us? Should we risk loss of control of our civilization?”
In response, the Association for the Advancement of Artificial Intelligence published its own letter citing the many positive differences that AI is already making in our lives and noting existing efforts to improve AI safety and to understand its impacts. Indeed, there are important ongoing gatherings about AI regulation like the Partnership on AI’s recent convening on Responsible Generative AI, which happened just this past week. The UK has already announced its intention to regulate AI, albeit with a light, “pro-innovation” touch. In the US, Senate Minority Leader Charles Schumer has announced plans to introduce “a framework that outlines a new regulatory regime” for AI. The EU is sure to follow, in the worst case leading to a patchwork of conflicting regulations.
All of these efforts reflect the general consensus that regulations should address issues like data privacy and ownership, bias and fairness, transparency, accountability, and standards. OpenAI’s own AI safety and responsibility guidelines cite those same goals, but in addition call out what many people consider the central, most general question: how do we align AI-based decisions with human values? They write:
“AI systems are becoming a part of everyday life. The key is to ensure that these machines are aligned with human intentions and values.”
But whose human values? Those of the benevolent idealists that most AI critics aspire to be? Those of a public company bound to put shareholder value ahead of customers, suppliers, and society as a whole? Those of criminals or rogue states bent on causing harm to others? Those of someone well meaning who, like Aladdin, expresses an ill-considered wish to an all-powerful AI genie?
There is no simple way to solve the alignment problem. But alignment will be impossible without robust institutions for disclosure and auditing. If we want prosocial outcomes, we need to design and report on the metrics that explicitly aim for those outcomes and measure the extent to which they have been achieved. That is a crucial first step, and we should take it immediately. These systems are still very much under human control. For now, at least, they do what they are told, and when the results don’t match expectations, their training is quickly improved. What we need to know is what they are being told.
News of the Week
China’s new GDP figures may restore faith in its economy
But it still has deep economic problems to overcome
Apr 18th 2023
To read more of The Economist’s data journalism visit our Graphic Detail page.
Since China abandoned its zero-covid regime at the end of last year, its economic recovery has been both hotly anticipated and closely scrutinised by investors. Sceptics had begun to wonder if demoralised consumers would open their wallets, if faith in a broken property market could be restored, and if a global downswing would undermine China’s upward momentum.
Official figures released on April 18th should help allay some of that scepticism. China’s GDP grew by 4.5% in the first three months of this year, compared with the same period in 2022. That does not sound particularly impressive by China’s standards. But to get there, the economy first had to climb out of the hole created by the ruinous lockdowns in Shanghai and elsewhere last spring (see chart). Much of China’s growth over the past 12 months was concentrated in the last three, during which it expanded at an annualised pace of about 9%.
Elon Musk wants to develop TruthGPT, ‘a maximum truth-seeking AI’
Ivan Mehta @indianidle / 5:50 AM PDT•April 18, 2023
Comment
Image Credits: CARINA JOHANSEN/NTB/AFP / Getty Images
On Tuesday, Elon Musk said in an interview with Fox News’ Tucker Carlson that he wants to develop his own chatbot called TruthGPT, which will be “a maximum truth-seeking AI” — whatever that means.
The Twitter owner said that he wants to create a third option to OpenAI and Google with an aim to “create more good than harm.”
“I’m going to start something which you call TruthGPT or a maximum truth-seeking AI that tries to understand the nature of the universe. And I think this might be the best path to safety in the sense that an AI that cares about understanding the universe is unlikely to annihilate humans because we are an interesting part of the universe,” Musk said during the Fox & Friends show.
Musk was also critical of OpenAI saying that he played a pivotal role in setting up the organization, but it’s not clear if it is doing any good. He also accused OpenAI of training AI models to be politically correct, which he considers “another way being untruthful.”
He previously criticized the company for becoming “closed sourced” and “effectively controlled by Microsoft” in a tweet in February.
OpenAI was created as an open source (which is why I named it “Open” AI), non-profit company to serve as a counterweight to Google, but now it has become a closed source, maximum-profit company effectively controlled by Microsoft.
Not what I intended at all.
— Elon Musk (@elonmusk) February 17, 2023
In March, Semafor reported that Musk and OpenAI co-founder Sam Altman had a tiff in 2018. When Musk wanted to take control of the company because it had “fallen fatally behind Google,” Altman and other co-founders objected to the move resulting in the Tesla CEO walking away.
Google Looks to Turbocharge AI Efforts With Combined Brain, DeepMind Unit
Search giant merges AI resources to accelerate progress in technology that is reshaping business
By Miles Kruppa
Updated April 20, 2023 5:15 pm ET
Tech companies including Google have been racing to build increasingly large AI algorithms. Photo: Robyn Beck/Agence France-Presse/Getty Images
Alphabet Inc.’s GOOG 0.84%increase; green up pointing triangle Google merged its two main artificial-intelligence research units, a major reshuffling as tech companies jockey for leadership in an area quickly reshaping business.
Google said the new unit, Google DeepMind, would combine the existing Brain and DeepMind research groups into one team. The unit will be led by Demis Hassabis, the co-founder and CEO of DeepMind, which Google purchased for about $500 million in 2014.
Jeff Dean, the previous head of Google Research overseeing the Brain team, will move to a new role as chief scientist. Google said he would lead the company’s “most critical and strategic technical projects related to AI,” including a new series of powerful AI programs.
“Combining all this talent into one focused team, backed by the computational resources of Google, will significantly accelerate our progress in AI,” Google CEO Sundar Pichai said Thursday in a blog post.
Tech companies including Google and Microsoft Corp.-backed MSFT -0.81%decrease; red down pointing triangle OpenAI, the creator of the popular chatbot ChatGPT, have been racing to build increasingly large AI algorithms known as large language models, or LLMs. Those programs have shown promise performing a range of tasks, such as summarizing complex documents and conducting humanlike conversations.
Google pioneered some of the technology that paved the way for recent developments in such models, but largely held back from wide public release of its most powerful image- and text-generation programs, citing the need to test the tools for accuracy and signs of bias.
VCs Challenged by the Downturn
David Cummings Entrepreneurship April 15, 2023
While current startup difficulties deserve the most attention (broken cap tables, valuations lower than capital raised, slowing growth rates, etc.), venture firms, especially ones that deployed significant amounts of capital in 2020 and 2021, are challenged as well. Writing checks into startups whereby their real valuations are likely 30 – 70% lower, even after a couple years of growing, makes for an exaggerated J curve (rates of return are negative in the first few years due to the drag from management fees). Only, the story is worse, and more complicated.
While venture fund fees are usually 2 and 20, meaning a 2% management fee and 20% of the carry (profits). The 2% management fee is paid on committed capital. With a $500M fund, that’s $10M/year in fees. Most funds are 10 years with two optional one year extensions, so best to think of the management fee side as really 20% of the fund (2% per year for 10 years). Instead of 2 and 20, 20 and 20 is the real way to think of it (20% of the fund as management fees and 20% of the profits as carried interest). Yes, the fees can be substantial, but it’s best to think of venture as get rich slowly, as most don’t beat the S&P benchmark and take 7+ years to see any distributions.
With venture funds, the general partners (GPs) are required to put 1-3% of the capital in themselves. So, if it’s a $500M fund, and the partners collectively are putting in 2%, they’re committing $10M. Many of the younger partners, while they’re making a good salary, often have to borrow money for their portion of the GP commit. Well, that debt they’ve taken on looks even more distressing considering it could be years before their portfolio company valuations are back to the original investment value, let alone at a scale where they’ll make good money on an exit.
Net net, the chance of this cohort of VCs making substantial profit is low yet personal debt is high. VC is a tough job to get, and right now, quite a few VCs are thinking about their current careers. Vintage 2020 and 2021 venture funds are likely to perform poorly save for the occasional power law winners. Venture is still an important part of our economy that has an outsized impact, and should be supported and encouraged, even if a couple years prove difficult.
Tiger Global Management’s $12.7 Billion Venture Fund Records 20% Loss
By Erin Woo
April 20, 2023 6:00 AM PDT
Tiger Global Management’s $12.7 billion venture fund, launched near the peak of the tech stock boom in October 2021, gave the New York hedge fund firepower to back hundreds of startups in under two years. Now the mammoth fund is bearing testimony to the bruising correction in private tech valuations that followed.
Tiger has told investors the fund had a paper loss of 20%, net of management fees, as of December, according to documents viewed by The Information. That’s worsened from an 8% loss it recorded as of June 2022 and an 11% loss as of September that year.
THE TAKEAWAY
Write-downs of crypto bets including OpenSea, Helium and FTX, as well as lowered valuations for some secondary investments in ByteDance, are dragging down returns for Tiger Global Management’s biggest venture fund.
The most recent declines reflect some total wipeouts in Tiger’s portfolio, the bankrupt crypto exchanges FTX and FTX US, as well as steep write-downs in other high-value crypto companies, such as non-fungible token marketplace OpenSea and Bored Ape Yacht Club, which produces NFTs. Tiger also wrote down the value of the fund’s investment in TikTok owner ByteDance, the documents show.
A spokesperson for Tiger declined to provide a comment.
Apple launches Apple Card’s savings accounts with 4.15% interest rate
Romain Dillet @romaindillet / 8:00 AM PDT•April 18, 2023
Image Credits: Apple
Apple Card customers in the U.S. can open a savings account and earn interest through an Apple saving account. To learn the specifics about Apple’s new offering, click here. When the company originally announced the new financial product back in October, Apple said that it couldn’t share what interest rate would be paid out on these accounts because rates are fluctuating so much these days.
As of today, Apple is going to offer an APY of 4.15%. It looks like a competitive offering when you look at data from Bankrate — you can currently find savings accounts that offer an APY of 3.5% to 4.75%. The company isn’t making any promise when it comes to future interest rates. It could go up and down at any time.
Apple has partnered with Goldman Sachs once again for the banking feature. Savings accounts are technically managed by Goldman Sachs, which means that balances are covered by the Federal Deposit Insurance Corporation (FDIC).
This high-yield savings account has been created specifically for Apple Card customers. When customers pay with their Apple Card, they get cash back on all purchases. By default, all purchases grant you 1% in cash rewards and 2% for all purchases made using Apple Pay. Purchases with select merchants unlock 3% in rewards.
Startup of the Week
SeatGeek Files Confidentially for IPO
April 19, 2023 12:34 PM PDT
Event ticketing company SeatGeek filed confidentially with regulators this month for an initial public offering, people familiar with the matter said, adding itself to a long list of firms aiming for IPOs once market conditions improve.
THE TAKEAWAY
SeatGeek has filed confidentially to go public, and it expects to pull in more than $500 million of revenue this year.
The New York-based company is hatching listing plans as a post-pandemic boom in live events helps bolster its business, which people use to buy and sell sports and concert tickets. SeatGeek, which makes most of its money on ticket transaction fees, expects to pull in more than $500 million in revenue this year, one of the people said. That would represent a roughly tripling of sales from 2021 that the company reported in a filing last year.
The filing with the Securities and Exchange Commission doesn’t guarantee that the company will go public this year, one of the people cautioned. There is a glut of companies looking to go public, especially those that have been operating privately for a decade or more, but market conditions haven’t been hospitable. Steep interest rate hikes and the impacts of the Silicon Valley Bank collapse have cooled most tech firms’ appetites to go public this year, said one IPO adviser working with several tech companies.
Tweet of the Week
@NeilThanedar
