Hitting The Bottom?
That Was The Week 2023, #12
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 my editorial and the weekly video.
Content this week from @geneteare, @crunchbasenews, @ttunguz @wilmerhale, @_RosieBradbury, @stevemollman, @CaseyNewton, @odsc, @pkedrosky @GaryMarcus, @SubstackInc, @bypatrickgeorge
Editorial: Are we at the bottom yet?
Essays of the Week
Global VC Funding Falls Dramatically Across All Stages In Rocky Q1, Despite Massive OpenAI And Stripe Deals Gené Teare
The Change in Burn Multiple for Startups in 2023 - Tomasz Tunguz
2023 IPO, Venture Capital and M&A Reports - Wilmer Hale
Investors dream big at Saudi-backed Florida Summit - Circuit News
Saudi Arabia's PIF reveals VC ties as influence grows - Rosie Bradbury
American investors shouldn’t be ‘arming the enemy’ by helping China create its own version of OpenAI, warns top VC Keith Rabois - Steve Mollman
Why journalists can't quit Twitter -
AI of the Week
The takeaways from Stanford’s 386-page report on the state of AI - Devin Coldewey
Is GPT-4 the Key to Unlocking Artificial General Intelligence? New Research Suggests So - OSDC
AI, Mass Evolution, and Weickian Loops
Humanness In The Age Of AI - Worldcoin
I am not afraid of robots. I am afraid of people. -
News Of the Week
Theory - Tomasz Tunguz
Introducing Substack Notes - Substack
Tiger Global Was Looking To Siphon Off VC Fund Stakes - Keerthi Vedantam
Richard Branson's Virgin Orbit Goes Bankrupt - Luc Olinga
Startup of the Week
YC Demo Day Part 1 and 2 - TechCrunch
Tweet of the Week
Patrick George @bypatrickgeorge
I woke up this morning to CNBC’s Squawk on the Street. The headline was from my wife Gené’s review of Q1 venture funding from Crunchbase News. Venture down 53%.
Ben Lerer from Lerer Hippeau, a New York early-stage venture fund, was in the chair fielding questions. He did a good job of confirming the gloomy outlook. Most investors this week are echoing the trend. It’s really bad out there.
I gave the Crinchbase article to ChatGPT4 and asked for a Blues song summing up its main points. Its really quite good:
(Verse 1) I woke up this morning, with the venture blues on my mind Yeah, I woke up this morning, with the venture blues on my mind Investors pullin' back, leavin' startups far behind
(Verse 2) Global funding's down, yeah, it's been a rough year Oh, the global funding's down, Lord, it's been a rough year With Silicon Valley Bank collapsin', oh, the end feels near
(Chorus) I woke up this morning, with the venture blues in my soul Oh, I woke up this morning, with the venture blues in my soul But we'll keep on pushin', till we reach our goals
(Verse 3) Seed funding's droppin', investors holdin' tight Yeah, the seed funding's droppin', investors holdin' tight But we'll remember Square and Airbnb, who found the light
(Verse 4) Early-stage funding's down, it's a 54% decline Oh, early-stage funding's down, a 54% decline But we keep on hopin', Lord, that we'll be just fine
(Chorus) I woke up this morning, with the venture blues in my soul Oh, I woke up this morning, with the venture blues in my soul But we'll keep on pushin', till we reach our goals
(Verse 5) Late-stage funding's fallin', but AI's still standing tall Oh, late-stage funding's fallin', but AI's still standing tall With OpenAI and Stripe, we'll rise above it all
(Verse 6) The slowdown's hit us hard, but we'll keep on movin' through Yeah, the slowdown's hit us hard, but we'll keep on movin' through With the venture blues in our hearts, there's nothing we can't do
(Chorus) I woke up this morning, with the venture blues in my soul Oh, I woke up this morning, with the venture blues in my soul But we'll keep on pushin', till we reach our goals
So what is going on?
Well, I may not be the best person to say. I’m pleased to say that SignalRank () closed its second round of funding last Friday. At least one well-known reader of this newsletter invested. The company is now ready to support seed and Series A investors in doing their pro-rata share of recommended Series B rounds.
The trend in Series B rounds is certainly down, as Tomasz Tunguz wrote a couple of weeks ago. Tomasz has a new $230m fund announced this week - Theory (see below). But great Series Bs are still happening.
That said, the decline is real, and the bottom is not in sight. The cause of the decline is not unlike a train derailment. A coach leaves the tracks - in this case, the decline in public market values driven by high inflation and interest rates. Then the IPO market disappears. After that, late-stage venture-backed companies struggle to raise capital at any price. Then Series B rounds decline in number and value. And now seed and Series A rounds are also derailed.
Only when public markets recover will the entire train be able to get back on the rails. Until then, venture-backed companies must preserve capital and align spending to revenue growth. The only exception is in the AI space, where large bets are being placed on huge outcomes.
Investors with capital are in a very good position. Great founders are still building startups that can grow into industry-defining pioneers. The investments are being done at lower valuations and raising less capital. Capital efficiency is now well understood. It is likely that Series A and B rounds completed in the next two to three years will produce the best outcomes for many years.
So we are not at the bottom, but we will likely get there over the next few months. And in the meantime, the next top is already being constructed. Smart investors know that to be true.
Essays of the Week
Global VC Funding Falls Dramatically Across All Stages In Rocky Q1, Despite Massive OpenAI And Stripe Deals
April 5, 2023
Venture and growth investors in private companies continued to scale back their investment pace in the first quarter of 2023, Crunchbase data shows.
Global funding in the first quarter reached $76 billion — marking a 53% decline year over year from $162 billion in the first quarter of 2022. That’s even including a reported $10 billion investment into OpenAI — largely from Microsoft — and a $6.5 billion round for payments giant Stripe. Without those two large deals, Q1 venture funding would have been down even more dramatically, close to $60 billion.
Every funding stage last quarter was down 44%-54% year over year, a clear signal that the slowdown is not confined to late-stage funding. Investors across each stage scaled back as they took time to assess new investment opportunities while guiding existing portfolio companies.
The collapse of Silicon Valley Bank on March 10 was an added shock to a weakened funding environment. The bank had more than 20,000 startup depositors with $5 million or less in revenue.
The impact was not limited to U.S. startups, either: SVB was also the bank of choice for startups around the world, sourcing funding from U.S. venture firms. Investors and founders scrambled to secure funds to meet payroll as a swath of startups faced possible closure through lack of funds.
The reset in 2022 has deepened across all funding stages quarter over quarter and year over year — outside of the blips from a few multibillion-dollar, late-stage financings. Sectors that were down in the first quarter include e-commerce, blockchain and cryptocurrency.
But artificial intelligence remains a bright spot, with large fundings in Q1 2023. Companies tagged with AI accounted for 19% of investment dollars last quarter, with the investments in OpenAI and hundreds of millions invested in the likes of Anthropic, SandboxAQ and Adept AI, according to Crunchbase data.
The slowdown is having a variable impact on different regions. More to come in the next week on how different geographies are impacted by the big reset in venture capital.
The data contained in this report comes directly from Crunchbase, and is based on reported data. Data reported is as of April 3, 2023.
Note that data lags are most pronounced at the earliest stages of venture activity, with seed funding amounts increasing significantly after the end of a quarter/year.
Please note that all funding values are given in U.S. dollars unless otherwise noted. Crunchbase converts foreign currencies to U.S. dollars at the prevailing spot rate from the date funding rounds, acquisitions, IPOs and other financial events are reported. Even if those events were added to Crunchbase long after the event was announced, foreign currency transactions are converted at the historic spot price.
The Change in Burn Multiple for Startups in 2023
Burn multiple measures the capital efficiency of a startup. Burn multiple calculated like this : net burn divided by net new ARR. Startup burn multiples have changed markedly in 2023.
Companies with more efficient burn multiples between 1-2 plan to increase their net burn per new dollar of bookings by between 14-40%. In contrast, startups with higher operating expenses will reduce their burn multiple by 34-70%. The average startup with a burn multiple of 5 plans to reduce to 1.5.
Companies with burn multiples of 3 will stand firm, operating their businesses at the same levels of efficiency.
Curiously, there’s no correlation between burn multiple & expected growth rate in 2023.
The majority of the surveyed population plans to operate with burn multiple of 1.5 or less.
With the Series B & Series C capital markets barely open, the quantum of growth capital has plummeted nearly 80%.
Startups’ plans have shifted aggressively to capital efficiency, without a noticeable impact on ARR growth. We’re all learning to achieve the same with less.
2023 IPO, Venture Capital and M&A Reports
MARCH 31, 2023
WilmerHale’s annual IPO, Venture Capital and M&A reports offer detailed insights into market conditions and provide comprehensive statistics and thoughtful analysis.
The 2023 IPO Report offers a detailed IPO market review and outlook, plus useful market metrics and need-to-know information for pre-IPO companies on the following topics:
Recent trends in the adoption of JOBS Act relief by emerging growth companies (EGCs).
What EGCs should consider as they prepare to exit EGC status.
The resurgence of the reverse merger—an IPO alternative gaining traction in the life sciences sector.
New SEC rule amendments governing the use of Rule 10b5-1 trading plans.
What IPO companies should know about blackout provisions in light of new SEC insider trading policy disclosure requirements.
How a change to the Delaware General Corporation Law is expanding liability protection for company officers.
The new SEC rule that will require compensation clawbacks for exchange-listed public companies.
Evolving business needs, investor expectations and regulations that are shaping board composition.
The 2023 Venture Capital Report offers an in-depth US venture capital market analysis and outlook, including industry and regional breakdowns, and insights on the following topics:
The implications of new beneficial ownership reporting requirements for private companies under the Corporate Transparency Act.
Challenges posed by the expanding patchwork of state salary disclosure laws.
How SEC safe harbors can help pre-IPO companies weather the rigors of the “quiet period.”
What investors need to know about state taxes on qualified small business stock.
Deal term trends in VC-backed M&A transactions and convertible note, SAFE and venture capital financings.
The 2023 M&A Report offers a global M&A market review and outlook, as well as insights on the following topics:
Trends in takeover defenses for public companies.
Recent cases that have placed Delaware corporations in a stronger position to limit shareholder books and records demands.
How deal terms compare in public and private company acquisitions.
Trends in VC-backed company M&A deal terms.
The relevant reports also showcase prominent recent capital markets, venture capital and M&A transactions.
Investors dream big at Saudi-backed Florida Summit
Adam Neumann admires desert kingdom’s readiness to `start from scratch’ as he builds residential real estate company Flow with Andreessen Horowitz
April 3, 2023
From WeWork founder Adam Neumann and former White House advisor Jared Kushner to LIV Golf Commissioner Greg Norman, a parade of investors and entrepreneurs mounted a stage in Miami Beach, Fla., last week to talk about why Saudi Arabia looms large in their future business plans.
Neumann, now building a residential real estate business called Flow that secured $350 million in funding from Silicon Valley venture capital firm Andreessen Horowitz, told the Future Investment Initiative’s Global Priority Summit on Friday that he sees the desert kingdom as an ideal environment for reinventing the way people live and work.
“What’s happening in Saudi right now is they’re dreaming up cities,” the 43-year-old Israeli who popularized the concept of shared office space said in an onstage conversation with investors Marc Andreessen and Ben Horowitz. “You’re starting from scratch. And that’s unbelievable, because you can apply the right technology and the right energy from the first day.”
Neumann, who also spoke about difficult personal lessons he drew after getting ousted as WeWork CEO in 2019, was among the most anticipated speakers at the two-day meeting sponsored by FII, a spinoff of the group’s annual economic conference in Riyadh, Saudi Arabia, that is colloquially known as Davos in the Desert. Both are backed by the kingdom’s Public Investment Fund, a $650 billion sovereign wealth fund whose investments run from crypto startups and gaming studios to the upstart LIV Golf league.
Yasir Al-Rumayyan, governor of PIF and chairman of state oil company Saudi Aramco, opened the proceedings on Thursday by saying he expects the PIF to grow to $1 trillion in assets by 2025 and between $2 trillion and $3 trillion by 2030. As head of Aramco, whose 2022 net income of $161 billion was the highest of any publicly traded company in the world, Al-Rumayyan said the company is committed to mitigating climate change but that shifting to sustainable energy sources needs to be gradual.
“Some of the governments around the world bullied the oil and gas companies,” Al-Rumayyan said. “It takes time to have a transition from fossil fuels to renewables.”
The Miami conference comes amid efforts by Americans and Saudis to ease tensions between the two countries, which have been close allies for more than seven decades.
Kushner, founder of Affinity Equity Partners, which has received a reported $2 billion in investment from the PIF, said he sees great opportunities in both the Saudi kingdom and the Florida city where he moved after departing the White House.
“I think Saudi Arabia and Miami are two places in the world that are shining examples of places that are on the rise, places that are not locked down by the past, places that are open-minded, [where] people are coming together,” he said.
Miami Mayor Francis X. Suarez, who opened the conference alongside Al-Rumayyan, said he learned about the opportunities in Saudi Arabia when he attended the three-day Riyadh meeting in October, which drew more than 6,000 participants, and agreed to host the Priority Summit in his city. Notable attendees included investors Nelson Peltz of Trian Fund Management, Steve Wynn of Wynn Resorts, Josh Harris of 26North and Barry Sternlicht of Starwood Capital.
Reflecting on the Abraham Accords, the 2020 set of agreements he spearheaded to normalize Israel’s relations with the United Arab Emirates, Bahrain, Morocco and Sudan, Kushner said he was gratified to see regional political changes leading to wider commercial opportunities.
“The practical implications have been tremendous right now,” he said. “You have business happening in the Middle East that would not have been… you have defense agreements that are happening, which wouldn’t happen… The biggest thing for me is really two things. Number one is you’re seeing Arabs now, finally, and Muslims being able to say nice things about Israel and Jews, and you’re having human-to-human connections, which I think is now finally pushing back against the opposition, which is hard.”
Saudi Arabia's PIF reveals VC ties as influence grows
April 4, 2023
The venture arm of PIF, Saudi Arabia's $650 billion sovereign wealth fund, has ramped up its dealmaking in recent years, but the extent of its influence in Silicon Valley was a matter of speculation—until now.
Sanabil Investments, which commits around $2 billion annually in VC, growth and buyout investments, recently publicized firms to which it has committed, as well as direct investments, on its website. The disclosures, first reported by The Information, reveal a greater scope of participation by PIF's venture arm in US startups and VCs than previously known.
The roster of firms spans dozens of US VC and PE investors, including Andreessen Horowitz, 500 Global, Coatue and KKR. The list of companies that Sanabil has backed includes e-scooter provider Bird, banking startup Varo and Oura, the maker of a wearable health-monitoring device.
The sudden burst of transparency follows a significant increase in startup investments by Sanabil in recent years. Founded in 2009, 22 of the wealth fund's 38 previously known investments took place in 2022, according to PitchBook data.
A majority of the direct investments listed on Sanabil's website did not have a previously known connection to the wealth fund's growth investment arm. Those companies were ActionIQ, Bird, CallSign, EMA, Oura, Varo, Vectra AI, consultancy firm Richard Attias & Associates and VC firm NEA. When contacted by PitchBook, Varo declined to comment, and the others did not respond.
Some investors and founders sought to distance themselves from the Saudi royal family following the 2018 killing of journalist Jamal Khashoggi. Even so, Sanabil, and parent PIF, have increasingly flexed their muscles on the global venture landscape. Some VCs appear to be actively courting them, despite long-standing criticism of Saudi Arabia's human rights record.
In a Miami conference panel Friday, Andreessen Horowitz co-founder and general partner Ben Horowitz reportedly praised Saudi Arabia as a "startup country."
American investors shouldn’t be ‘arming the enemy’ by helping China create its own version of OpenAI, warns top VC Keith Rabois
April 6, 2023 at 4:53 PM PDT
Keith Rabois, general partner of Founders Fund.
DAVID PAUL MORRIS—BLOOMBERG VIA GETTY IMAGES
OpenAI’s ChatGPT may have taken the world by storm, but the A.I. chatbot is blocked in China, as are many internet apps, including Facebook and YouTube. Predictably, Chinese startups are racing to become their nation’s version of Microsoft-backed OpenAI.
American money is indirectly helping them out, and Keith Rabois, a general partner at venture capital firm Founders Fund, has a problem with that.
On Wednesday, the PayPal Mafia alum tweeted, “This needs to be illegal” while sharing an article from tech news site The Information entitled “Sequoia and Other U.S.-Backed VCs Are Funding China’s Answer to OpenAI.”
The article outlines how American institutional investors, including U.S. endowments, back Chinese VC firms that in turn are investing in Chinese A.I. startups. Among those firms is Sequoia Capital China, the Chinese affiliate of the Silicon Valley VC giant.
When another Twitter user noted the VCs were technically Chinese VC firms, not U.S. ones—albeit with some U.S. ties—Rabois added the firms “should not be allowed to take any LP [limited partner] money from the US.”
The Biden administration is reportedly mulling an executive order, with national security risks in mind, that would impose new controls on U.S. investors looking to support Chinese projects on certain technologies, including semiconductors, and A.I. Rabois suggested Wednesday that the White House should “move on it already.”
Rabois added in another follow-up tweet that “investing in arming the enemy” should be illegal.
According to The Information, Sequoia China recently made a U.S.-dollar investment in a new A.I. startup led by a former Google employee who has published research related to large language models, similar to the kind developed by OpenAI and Google and used for their A.I. chatbots.
Of course, there’s nothing new about Chinese tech startups raising money in U.S. dollars from VC funds that have backing from U.S. pension funds or university endowments.
But the power of A.I. systems like OpenAI’s ChatGPT—and the more capable successor GPT-4—has alarmed many observers. Last month, Tesla CEO Elon Musk and Apple cofounder Steve Wozniak were among a large group of experts calling for a six-month pause on developing any A.I. system more powerful than GPT-4.
Why journalists can't quit Twitter
The media should be building alternatives. Instead, some are doubling down
Today, Donald Trump was charged with 34 felony counts of falsifying business records in connection with a series of hush-money payments related to the 2016 US presidential campaign. His arraignment was carried live on cable news and National Public radio, but I learned of the day’s events where I still see almost everything first: Twitter, which, despite its perilous decline under Elon Musk, remains home base for the U.S. press corps even as the site itself increasingly orients itself to make fools of them.
In December, I predicted that 2023 would be the year that the media would begin its divorce from Twitter. “Elon Musk’s continued promotion of right-wing causes and personalities will push away more and more high-profile users, who find themselves increasingly put off by his shock-jock antics and whim-based approach to content moderation,” I wrote. “Alternative platforms like Mastodon, while smaller and less intuitive to use, offer a safe haven to more and more people — particularly journalists — looking for off-ramps. By the end of 2023, Twitter no longer sets the daily news agenda by default for the entire US press.”
Almost four months later, this prediction looks more and more wobbly. The first part has more or less come true: journalists are put off by Musk’s antics, and dunk on him daily. But those same journalists — along with a bunch of people Musk arbitrarily suspended, fired, or laid off — continue to tweet just the same, propping up the service with their quips and sports tweets and food photos just as they always have. And while some of the company’s competitors show intermittent signs on life, none has taken on the feeling of a daily must-visit in the way Twitter did and still does.
By some measures, Twitter usage is down from previous highs. But among the 900 or so tech and media professionals I follow, usage is basically steady. The timeline may scroll a little more slowly than it used to, but everyone is still showing up for their daily dose of sparring for retweets. And I think it’s worth taking a moment to reflect on why.
The first and most obvious reason is inertia. Journalists spent more than a decade building up their presences on Twitter, and they were never going to abandon the site collectively overnight. As long as they can still drive traffic to their stories, discuss those stories in public with their peers, and grow their audiences, they have little reason to leave. Some, like Big Technology’s Alex Kantrowitz, even see opportunity amid the decline: if big publishers won’t pay for Twitter Blue and the potentially bigger audience that comes with it, he will — and, he reasons, his own work will stand out more.
The second answer, I think, lies in the power of chaos to command our attention. Just as Trump’s tumultuous White House constantly kept the news cycle off balance, so, too, does Musk’s erratic leadership of reporters’ favorite website. On Monday, everyone logged on to see that the old birdhouse logo on Twitter’s website had been mysteriously replaced with the Shiba Inu from the doge memes. Was it an April Fool’s Day joke that shipped a little late? A wink at a lawsuit Musk faces over charges that he unlawfully promoted the Dogecoin currency? Or just something to make the boss laugh?
Whoever might know the answer isn’t saying. But the mystery has become part of Twitter 2.0’s allure. Will reporters and news outlets lose their verification badges? (Yes, but only at random.) Will the site even work? (Yes, but less than it used to.) What new view counts will be added to each tweet? (Here, for no particular reason, is the bookmark view count.) Where else will Musk’s name be found in the site’s source code? (In the recommendation algorithm, ensuring his tweets are seen more than other people’s.)
Compare this Lost-style alternate reality game with the humdrum boredom of Twitter’s upstart rivals: Mastodon, T2, Post. Each has something to recommend it — I’ve been spending most of my posting time lately on Mastodon, whose low-drama founder and decentralized structure appeal to me — but all lack the animating force of Musk as a main character, rampaging across the timeline each day like a Tesla-driving Green Goblin.
Finally, the alternative apps are all still missing something. Mastodon’s decentralized network can make basic concepts like following another account more difficult than users are used to; its lack of an equivalent of quote-tweeting (for now) makes a lot of basic Twitter-like discourse impossible. Post and T2 lack original content or distinguishing features. Artifact hasn’t shipped its social feed out of beta yet.
I find all of this depressing. In December I began posting to Twitter only as a kind of RSS feed, re-sharing my Mastodon posts about new newsletter editions and podcasts, hoping that others would join me. To date, few have. I miss the conversations I used to have on Twitter, but not enough to contribute free labor there.
It may be that Twitter is immortal after all. But there are still more shoes yet to drop. There will almost certainly be a series of regulatory fines related to hate speech and data privacy, potentially reaching into the billions of dollars. The collapse in advertising and API revenue, mounting debt payments, and the utter flop that has been Twitter Blue could all lead the site into bankruptcy. And should Musk follow through with plans to make the site basically unusable for free accounts, Mastodon and its peers will have a renewed opportunity to attract users.
For the moment, though, Musk has learned the same lesson Jack Dorsey did: Twitter is extremely hard to kill. And for the journalists who have come to rely on it, there is almost no indignity they won’t suffer to get their fix.
The other day, the company’s old press email address started responding to reporters’ queries with an automated 💩 emoji. Musk tweeted about it, and reporters soon gleefully confirmed it with screenshots of their own. The company was symbolically shitting all over them, and journalists couldn’t get enough of it.
AI of the Week
The takeaways from Stanford’s 386-page report on the state of AI
Devin Coldewey @techcrunch / 11:17 AM PDT•April 4, 2023
Image Credits: Getty Images
Writing a report on the state of AI must feel a lot like building on shifting sands: By the time you hit publish, the whole industry has changed under your feet. But there are still important trends and takeaways in Stanford’s 386-page bid to summarize this complex and fast-moving domain.
The AI Index, from the Institute for Human-Centered Artificial Intelligence, worked with experts from academia and private industry to collect information and predictions on the matter. As a yearly effort (and by the size of it, you can bet they’re already hard at work laying out the next one), this may not be the freshest take on AI, but these periodic broad surveys are important to keep one’s finger on the pulse of industry.
This year’s report includes “new analysis on foundation models, including their geopolitics and training costs, the environmental impact of AI systems, K-12 AI education, and public opinion trends in AI,” plus a look at policy in a hundred new countries.
Let us just bullet the highest-level takeaways:
AI development has flipped over the last decade from academia-led to industry-led, by a large margin, and this shows no sign of changing.
It’s becoming difficult to test models on traditional benchmarks and a new paradigm may be needed here.
The energy footprint of AI training and use is becoming considerable, but we have yet to see how it may add efficiencies elsewhere.
The number of “AI incidents and controversies” has increased by a factor of 26 since 2012, which actually seems a bit low.
AI-related skills and job postings are increasing, but not as fast as you’d think.
Policymakers, however, are falling over themselves trying to write a definitive AI bill, a fool’s errand if there ever was one.
Investment has temporarily stalled, but that’s after an astronomic increase over the last decade.
More than 70% of Chinese, Saudi, and Indian respondents felt AI had more benefits than drawbacks. Americans? 35%.
But the report goes into detail on many topics and subtopics and is quite readable and nontechnical. Only the dedicated will read all 386 pages of analysis, but really, just about any motivated body could….
Is GPT-4 the Key to Unlocking Artificial General Intelligence? New Research Suggests So
After conducting 154 pages of tests, a recent paper has concluded that GPT-4 could potentially be an early version of an artificial general intelligence system. For those who don’t know, this would be a major development as an AGI system would be able to mimic the human mind by reasoning, learning, and problem-solving. Though the research has sparked interest in the data science community as many experts working with these systems seem to believe that GPT-4 represents something extraordinary, it doesn’t exactly mean what science fiction has popularized.
As you’d imagine, the paper highlights the breadth and depth of GPT-4’s capabilities. And its conclusion suggests that GPT-4 may be viewed as an incomplete version of artificial general intelligence. While the findings should be taken with skepticism, the confidence of those involved in its development is noteworthy. That’s because many within the AI field consider AGI as the ultimate goal of their research.
OpenAI, the company behind ChatGPT, is among those at the forefront of the development of artificial general intelligence. The company’s CEO, Sam Altman has expressed confidence in their ability to build AGI, making it clear that they are all-in on the project. He also took to Twitter to express how artificial general intelligence systems could help solve many of humanity’s problems last summer:
AI, Mass Evolution, and Weickian Loops
The early stages of technology development around a new platform tend to be simple wrappers. Back when microprocessors emerged we saw single-board computers, simple wrappers that abstracted away some of the complexity; when personal computer operating systems arrived, some of the first applications were just wrappers on operating system functions themselves, like Norton Utilities.
We have seen this play out over and over on the Internet itself. Many of the most popular early applications were just wrappers on existing Internet and Unix applications. Mail wrapped around Unix mail; Twitter can be thought of as a variant of the Unix function “finger” for .plan files. Simple wrappers are often very popular and can be financially successful, so being a wrapper is not pejorative. Rather, it’s just that as ecosystems evolve to exploit new platforms, the preponderance of new and widely-used applications tends to evolve away from these wrappers.
Application evolution itself changed, of course, with Internet apps and services. For the first time, there was a more direct two-way conduit between what users did with applications, and what developers could see about what they were doing. This led to a more coupled system, and more rapid application development, where many features were created in reaction to real data about how products were used, as opposed to instincts, or interviews, or surveys, all of which were popular before it became possible to get that information directly.
Over time, as developers get a better sense of what is possible, and as users evolve new needs based on what they see, applications tend to change dramatically, becoming more sophisticated, and less like simple wrappers. The iterative process continues and can be slow, but the trend remains, that of abstracting away from simple wrappers toward more complex applications, fueled by user data. And, at the same, the data flows and loops more seamlessly from usage back to development, leading to more directed enhancements.
We see all of this with the emergence of a new class of AI applications, often built on top of chat AIs, generative AI, and so on. The first few attention-getting applications and services have largely been wrappers, whether around marketing (click here to build a blog post for teenagers promoting XYZ), image creation (here is a prompt to create pictures of happy skiers), or office activities, like spreadsheets and presentations (create a six-slide deck for a typical hardware product go-to-market plan).
At the same time, the feedback loops are changing. No longer asynchronous, like in the PC era, or narrowly synchronous, like in the Internet era, the feedback is now fully-duplex and low latency synchronous. You can see this in the Discord server of the popular and controversial generative AI service Midjourney, where people can see one another’s image creation prompts, and the results, and then they rapidly iterate in realtime on top of each other’s prompts, moving rapidly to fringe ideas, as well as much higher levels of sophistication and verisimilitude. You are seeing, for the first time, real-time, mass evolution, as people and groups are de facto creating increasingly complex applications right in front of you.
These two changes—the shift away from wrappers to more complex and richer applications and services, and the increasingly fully duplex nature of the evolutionary pressuring forces—changes the nature of this phase of AI app development. It is a futurist cliché, but the pace must be swifter than ever, given the ease of iterating on spoken and written language, as opposed to the formal grammars of closed programming languages and compiled applications. It is faster and more democratic, given the collapse of de facto coding costs, as well as the disappearance of just year-ago barriers to participating in this iterative process.
This leads to interesting and important questions. For example, and we can push this analogy too far, we know that in biology over-fast evolution leads to instabilities; we know that slow-evolving species tend to do better than fast-evolving ones, in part because the latter respond too readily to transient stimulus, rather than exploiting their ecological niche.
HUMANNESS IN THE AGE OF AI
A path to an open and permissionless identity protocol.
The Worldcoin project is initiating an open and permissionless identity protocol called World ID. It empowers individuals to verify their humanness online while maintaining their anonymity through zero-knowledge proofs. Advancements in AI make it difficult to distinguish between AI and humans on the internet, highlighting a need for authentic human recognition and verification. To help address this, individuals can receive a future-proof unique human credential through a secure biometric device in a privacy-preserving manner. Importantly, this enables a path to AI-funded non-state UBI and the equitable global distribution of digital currencies. The project's early contributors have developed initial versions of a hardware device, a mobile client, and a deployment mechanism—all of which will gradually become decentralized. World ID will ultimately enable the protocol to be governed by the people, ensuring that the protocol benefits everyone.
The capabilities of AI are rapidly approaching those of humans and have already surpassed them in many niche areas. While the recent rise of Large Language Models (LLMs) demonstrates that these models are becoming increasingly versatile and seemingly also more “generally intelligent”, they feel a lot more intelligent because they perfected learning the primary interface with humans: language. Although it’s not immediately evident that this breakthrough will lead to Artificial General Intelligence (AGI) in the short-term, we now have models that are trained to perfectly mimic highly capable humans in digital interactions. Some might argue that the age of AI has begun.
The performance of deep learning models is improving at an accelerating pace, reaching superhuman levels in benchmarks with ever-increasing speed. Language models have recently experienced significant advancements, attaining performance within the upper 20th percentile of human capabilities on the majority of conventional assessments. When comparing GPT-3.5 to GPT-4, separated by only a few months, the rate of progress becomes especially clear.
Access to increasingly powerful models is becoming available to individuals in ways that are impossible to control. The Stable Diffusion image generation model and software to generate deep fakes are open source, and Meta’s LLaMA language model has been leaked and can be run on a laptop.
A significant short-term consequence is that, until recently, we deemed text-based messages sufficient for establishing and proving humanness—the famous Turing Test. Bots have been prevalent on social media platforms and involved in manipulation campaigns for quite some time. However, in most instances, it was feasible to differentiate them from authentic human users. Now, modern AI either has already passed the Turing Test or is very close to doing so. This will make it impossible in the future to determine humanness based solely on intelligence. Furthermore, recent impersonations using deep fakes have demonstrated that even video-based attestation of humanness is becoming increasingly unreliable. Consequently, there is no longer a reliable method to verify humanness online.
However, proving humanness in the digital domain will be an essential and likely inevitable tool for empowering individuals, especially in this new chapter of human history. While there are various approaches to how a proof of personhood (PoP) might end up being implemented, it is crucial that such an important infrastructure prioritizes privacy, self-sovereignty, inclusivity and decentralization in order to benefit and protect individuals.
In response to this challenge, the Worldcoin project has initiated an open identity protocol called World ID. The advancement of AI has rendered data that can be digitally generated insufficient for attesting one’s human status. As a result, the focus needs to shift toward verifying humanness through real-world attestations. To evaluate human status based on unique physical characteristics (i.e. biometrics), Tools for Humanity—a technology company contributing to the Worldcoin project—has supported the design of a custom, open-source hardware device. The state-of-the-art device, which uses several neural networks to validate liveness and uniqueness without storing any image data, issues an AI-safe PoP credential on World ID. While there are many ways to attest humanness (and different applications have different requirements), the current state of technology suggests that this approach is the only scalable, fraud resistant and inclusive mechanism for establishing global PoP.
I am not afraid of robots. I am afraid of people.
Some thoughts on AI risks, near-term and long-term, some recent controversies in AI, and why we are in trouble if we can’t find a way to work together
I’m scared. Most people I’ve spoken to in the AI community in the last few weeks are. I’m more concerned about bad outcomes for AI now than at any other point in my life. Many of us who have worked on AI are either feeling remorse or concerned that we might soon come to feel remorse; Ezra Klein wrote about a delegation of people from big AI companies speaking to him about their off-record-fears. I am putting my fear here, on the record.
Probably each of us is different. Some are frightened about some eventual superintelligence that might take over the world; my current fears have less to do with recent advances in technology per se, and more to do with recent observations about people.
I wasn’t feeling this a way year ago. Sure, I was skeptical that large language models would lead to general intelligence, I was worried about the coming thread of misinformation, and I was worried that we were wasting time on what LeCun later described as an off-ramp, but I was still seeing the AI community writ large as growing up, increasingly recognizing the value of responsible AI. Forward-thinking conferences like The ACM’s FAccT (Fairness, Accountability, and Transparency) were growing in size and influence. Corporations like Microsoft were putting out statements on responsible AI; I thought they had learned their lesson from the Tay debacle. Google was holding off on releasing LaMDA, knowing it wasn’t fully ready for prime time. On a 1-10 scale of least worried to most worried, I was at a 6. Now I am at an I- don’t-think-about-much-else 9.
Greed seemed to increase in late November, when ChatGPT started to take off. $ signs flashed. Microsoft started losing cloud market, and saw OpenAI’s work as a way to take back search. Nadella started to seem like a villain out of a Marvel movie, declaring in February he wants to make Google “come out and show that they can dance”, and “I want people to know that we made them dance.”
OpenAI, meanwhile, more or less finalized its transformation from its original mission
to a company that at present for all practical purposes is a for-profit, heavily constrained by the need to generate a financial return, working around Microsoft’s goals, with far less emphasis on the admirable humanitarian goals in their initial charter. Their 10 billion dollar deal earlier this year accelerated that transition.
Then along came Kevin Roose’s infamous batshit crazy mid-February conversation with Sydney; I thought for sure Microsoft would take Sydney down after that. But they didn’t. Instead they put up some risible band-aids (like limiting the lengths of conversations) and declared on 60 Minutes (same early March episode in which I had appeared) that they had solved the problems in 24 hours. The reality is far more complex. (E.g. one can still easily jailbreak Bing and get it to produce egregious misinformation.)
The real reason I signed The Letter was not because I thought the moratorium had any realistic chance of being implemented but because I thought it was time for people to speak up in a coordinated way.
At some level the petition worked; Capitol Hill, for example, pretty clearly got the message that we can’t sit around forever on these issues. But in other ways, I have even more concerns than before.
The main thing that worries me is that I see self-interest on all sides, and not just the corporate players, when what we really need is coordinated action.
What we need coordinated action on is actually pretty obvious, and it was spelled out perfectly clearly in the letter. Contrary to rumor, the letter didn’t call for a ban on research; it called for a shift in research.
AI research and development should be refocused on making today's powerful, state-of-the-art systems more accurate, safe, interpretable, transparent, robust, aligned, trustworthy, and loyal./
Who on earth could actually object to that? Whether or not you want a pause, the reality is that large language models remain exactly what I have always described them to be: bulls in a china shop, powerful, reckless, and difficult to control. Making new systems that are “more accurate, safe, interpretable, transparent, robust, aligned, trustworthy, and loyal” should literally be at the top of everyone’s agenda.
No sane person should have opposed that prescription.
News of the Week
Venture Capitalist at Theory
I’m thrilled to announce the debut of Theory Ventures & our first fund of $230m.
I’ve spent my career as a student of startups. I’ve worked for two, invested in many, seven of which became unicorns. I study them, benchmark them, analyze them, interview their leaders to understand their mechanics & share what I’ve learned on this blog.
The throughline of my work has been in identifying technological discontinuities that enable go-to-market advantages, & it’s patently clear to me that we are in the midst of another revolution now.
I aim to continue this work for many years at Theory Ventures.
Theory is deliberately named. Thesis investing is at our core. We aim to research ideas, develop informed perspectives, & ply those insights to support founders from their earliest stages.
These are the three of the first key thesis areas for Theory :
The Decade of Data: We are living in a decade of data. Every company leverages insight from data for competitive advantage. Data movement, transformation, analysis, & observability software will underpin data applications used by every part of modern organizations.
Machine Learning as a Force Multiplier: There are four types of machine learning: classification, prediction, interpretation, & generation. Modern software embeds these four types of ML into workflows which anticipate user needs & enable workers to operate at a superior level of abstraction.
Decentralized Infrastructure as Database: Blockchain technologies invert data ownership by shifting control to the end user. This new architecture transforms the relationship amongst users, assets, & businesses.
If you’re building in these spaces, or have a theory about the future of software, I’d love to hear from you.
Introducing Substack Notes
Unlocking the power of the subscription network
Hamish McKenzie, Chris Best, and Jairaj Sethi
We started Substack in 2017 because we wanted the internet to be better for writers and readers. We were dismayed with the clickbait and content farms, the listicles and liars, the cheap outrage and culture wars. We thought there could be something better if writers and readers were given more control and treated as a higher priority than advertisers, and if culture makers could find financial dignity without needing to sublimate themselves to attention games and corporate marketing budgets. “We believe that what you read matters,” we said, and we meant it.
So we set about building a system that fosters deep connections and quality over shallow engagement and dopamine hacks. We turned away from advertising and the attention economy and toward subscriptions and direct relationships. We believed that something beautiful could emerge from marrying the internet’s powers with a business model that puts writers and readers in charge, that rewards great work with money, and that protects the free press and free speech. We came to believe that this model—a subscription network—could form the basis of a new economic engine for culture.
There are more than 35 million active subscriptions to writers on Substack, including more than 2 million paid subscriptions. Readers have paid hundreds of millions of dollars to writers on the platform. There has been a Cambrian explosion of great writing, and writers have been saying (unprompted, we promise) that Substack has changed their lives. Encouraged by this early progress, we’ve become excited by the prospect of pushing the subscription network into new territory.
Notes: Post short-form content, recommend anything
In the coming days, we will start rolling out a way for writing, ideas, and discussion to travel through the Substack network. We’re calling this new product Notes.
In Notes, writers will be able to post short-form content and share ideas with each other and their readers. Like our Recommendations feature, Notes is designed to drive discovery across Substack. But while Recommendations lets writers promote publications, Notes will give them the ability to recommend almost anything—including posts, quotes, comments, images, and links. Our goal is to foster conversations that inspire, enlighten, and entertain, while giving writers a powerful growth channel as these interactions find new audiences.
Tiger Global Was Looking To Siphon Off VC Fund Stakes
April 6, 2023
It looks like one of the most prolific investors in tech is feeling less confident about its pandemic-era investments.
Tiger Global Management has been working with banks to sell off VC fund investments in the secondary market, The Information reported. It’s unclear what stakes the firm tried to sell, or if it was successful.
Tiger Global was an incredibly active investor in the tech ecosystem during the pandemic-era funding boom, and it also contributed investments to several VC firms during this time period. According to Crunchbase data, the firm invested in 139 startups in the first quarter of 2022 alone.
But following a dramatic dip in the stock market in 2022, and its ripples in the private market, Tiger Global reported in fundraising documents that its portfolio companies slashed their valuations by billions. Tiger Global, one of the largest investors in tech at the time, quickly stopped its investment streak. Per Crunchbase data, the company only invested in 45 startups in the latter half of 2022.
A bleak private market ahead?
It’s a clear sign that Tiger Global, and perhaps the venture community at large, isn’t confident in a market bounce back anytime soon. Tiger Global was one of the first wealth management companies to pull back from dealmaking back in February 2022. Indeed, global venture funding continues to slow down in Q1 2023, save for the tech world’s recent obsession with generative artificial intelligence.
But it looks like things are looking up for the company, albeit only slightly. Tiger Global’s hedge fund posted 7.3% gains in the first quarter of 2023, according to Bloomberg. The firm’s fund dropped 56% and its long-only fund dropped 67% between 2021 and 2022 (to recoup losses, each fund would need to see a gain of 144% and 216% respectively).
Richard Branson's Virgin Orbit Goes Bankrupt
The rocket company founded in 2017 by the British billionaire is desperately looking for a buyer.
APR 4, 2023 8:45 AM EDT
It's a sad day for the British aviation and space industry.
A few months ago, the country's ambitions to become a space power seemed to be coming true.
But in January, those hopes took a huge hit. Virgin Orbit, the rocket company founded in 2017 by knighted billionaire Sir Richard Branson, had decided to carry out the first rocket launch into space from the British soil.
The attempt was organized in collaboration with the British Space Agency and Spaceport Cornwall. Virgin Orbit wanted to launch nine satellites into space, which would have been a major first for the UK.
But the attempt was a resounding failure: An "anomaly” prevented the rocket from being put into orbit.
"The data is indicating that from the beginning of the second stage first burn, a fuel filter within the fuel feedline had been dislodged from its normal position," Virgin Orbit explained mid-February. "Additional data shows that the fuel pump that is downstream of the filter operated at a degraded efficiency level, resulting in the Newton 4 engine being starved for fuel. Performing in this anomalous manner resulted in the engine operating at a significantly higher than rated engine temperature."
'What Is Best for the Business'
The return to earth has been brutal since.
Virgin Orbit (VORB) , which is based in California, just filed for Chapter 11 bankruptcy, which means it can continue to do business while restructuring its debts, "in order to effectuate a sale of the business," the company said in a statement on April 4.
The firm said it intends to use the Chapter 11 process to maximize value for its business and assets.
"The team at Virgin Orbit has developed and brought into operation a new and innovative method of launching satellites into orbit, introducing new technology and managing great challenges and great risks along the way," Dan Hart, Chief Executive of Virgin Orbit, said. "While we have taken great efforts to address our financial position and secure additional financing, we ultimately must do what is best for the business."
The Chief Executive Officer added that: "We believe that the cutting-edge launch technology that this team has created will have wide appeal to buyers as we continue in the process to sell the company."
"At this stage, we believe that the Chapter 11 process represents the best path forward to identify and finalize an efficient and value-maximizing sale."
Virgin Orbit will receive $31.6 million from Virgin Investments, one of its sister companies, to stay afloat while it's looking for a buyer.
Startup of the Week
Our favorite startups from YC’s Winter 2023 Demo Day — Part 1
Comics software, meat-based plants and Tesla spun-out heat pumps, anyone?
Natasha Mascarenhas, Alex Wilhelm, Devin Coldewey, Anna Heim, Tim De Chant, Christine Hall/ 2:01 PM PDT•April 5, 2023
Image Credits: Bryce Durbin / TechCrunch
Here’s a crazy statistic: More than half of the companies in Y Combinator’s latest cohort were accepted to the accelerator with only an idea, no minimum viable product or revenue strategy needed.
Fast forward to today, where we’ll see how the individuals behind those ideas are thinking about the future of how we work, eat, connect and build. YC Demo Day is always a window into a subset of entrepreneurs, used to help us better understand how this world may play into the kinds of companies entering the market.
Before we get into some standout companies, here’s some background on the latest batch: The most common verticals were developer operations, open source and artificial intelligence. On the diversity front, YC says that 17% of the Winter 2023 companies have a woman founder, which is up from 15% in the previous batch.
OK, let’s start! Now, there’s a reason we are journalists, not investors, so let’s be clear: What we’ve written on the startups listed below are not meant as advice to join or back a startup; we’re highlighting which companies stood out to us during the rapid-fire pitching of Y Combinator’s 36th Demo Day.
Our favorite startups from YC’s Winter 2023 Demo Day — Part 2
Natasha Mascarenhas, Devin Coldewey, Anna Heim, Alex Wilhelm, Christine Hall, Tim De Chant, Karyne Levy/ 1:24 PM PDT•April 6, 2023
Over 20,000 applications flew into Y Combinator, which ended up plucking out 282 startups for its latest batch. And now we’re getting our first look at them through Demo Day.
The first day’s demos included a healthy dose of artificial intelligence and open source, which is different from years past that were dominated by new fintech companies. That’s not entirely surprising; we were less than 10 minutes into Demo Day before we heard the phrase “Cerebral Valley.” Our favorites from Wednesday also extend to EV charger credit cards and a Snowflake for sensor data.
Now we’re focused on the standouts from day two. Is crypto back? What happened to the accountant tech stack? And are we ready to start talking about cloud marketplaces? Check out our favorites from the second day of demos below and see for yourself.
And please remember that we’re not offering investing advice or recommending anyone join or back a startup. We’re just having fun.