WP Pro Online

Earlier today, we talked with Phil Haslett, the cofounder and now chief strategy officer of EquityZen, a 10-year-old, New York-based secondary marketplace that connects accredited buyers with privately held company shares that their owners — including founders, employees, and VCs — are looking to sell.

It’s a tough business to be running right now, competing as it is with shares of publicly traded companies that are selling at fire-sale prices compared with a year ago and are far more liquid. Indeed, like a lot of outfits, EquityZen last month conducted a sizable layoff, parting ways with 27% of its then 110-person team.

Still, Haslett believes adamantly that the secondary market will only grow bigger over time . . once it gets over this very big hump. More on what he’s seeing on pricing, hot and cold sectors, and more follows below in excerpts from our chat, lightly edited for length.

TC: The market was completely stuck back in June, with tons of demand to sell secondary shares but not a lot of buyers as people sat on the sidelines to figure out how bad things would get. What’s happening right now?

PH: The markets were pretty stagnant from April to maybe July or August owing to a combination of factors, the biggest being the pricing expectations sellers had where buyers really wanted to get into names. I’ve certainly seen an uptick. Mainly, I think what we’ve seen is reality setting in for selling shareholders on prices and also more buyers coming to the secondary space to find investments in names they like, because primary raises aren’t happening at all. If you’ve got a lot of capital to deploy, and you want to invest in late-stage tech, [and founders aren’t prepared to raise primary rounds] at a 40% discount to their last funding round, [investors] cross into the secondary space.

Yet you’re competing with publicly traded companies that are also very steeply discounted right now. In terms of transaction volume, how does it compare with a year ago?

I think any secondary platform or market participant would tell you that 2021 was a unique era for secondaries; probably no one is coming close to doing the amount of volume they did last year. [You’re right that if] you’re an investor, you might say, ‘There’s a really liquid solution out there where I can buy companies that are five times or even three times revenue in the public markets, so why would I enter into the private space?’ But once you’ve exhausted those opportunities, [the question becomes]: which are the private company names that you really still have a long-term belief in? And how can I as an investor deploy capital into those companies?

What are the ‘hottest’ brands on your platform right now?

Unfortunately, I can’t share actual names if you’re curious about sectors that are the most prominent, up until Q2, we were pretty active in web3 and crypto companies; that’s obviously gone really quiet of late. Fintech has retreated relative to last year. A consistent sector has been in cybersecurity; public names companies like CrowdStrike and Sentinel One and Zscaler and Palo Alto Networks have performed really well and that kind of feeds down into the private space where there are a lot of well-capitalized private companies that are solving a cybersecurity solution. Enterprise SaaS companies are still doing well, but [selling based] on a much more conservative multiple on revenue than in the past.

Are you seeing shares restricted by companies that don’t want it getting out that their secondary shares are selling at a huge discount to their last known valuation?

We’ve seen a bit of the opposite, which sounds counterintuitive, but you’ve got two opposing forces: venture capital firms and founders may be hesitant to have an active market that shows prices have gone down offset by employees and early investors who were thinking about a liquidity event this year or next year by way of an IPO and who’ve been completely shut out but have cash needs that are independent of the company’s performance. Also, when a story comes out like that of DataRobot, where a team of senior leaders got a bunch of liquidity when things were great and they didn’t extend that out to employees [who are dealing with the current market], that’s a complete egg on your face.

You work with a lot of founders and employees. Do you also handle institutional type trades? If a VC wants to sell a percentage of their entire portfolio to another buyer, can you handle that?

We do work with institutions; we work with venture capital firms that are buyers and sellers. I would say the trend that we’ve seen so far this year is seed stage funds that have some positions in their portfolio that have done tremendously well for them and are marked up and probably could return the entire value of the fund [ and they’re liquidating] some of that position so that they can return capital to their to their LPs. If you’re a seed stage fund to try to raise a new fund with no realized gains, that’s a tough conversation. Now, do they wish they had [sold a portion of those holdings] last year? I’m sure they do.

Of course, no one wants to catch a falling knife. Have you seen a bounce back at all in prices or are things still trending down?

Current average discounts to the previous funding round we’ve seen right now are at about 40%, which is the lowest we’ve seen. In Q1, it was probably closer to 20%. It’s name-specific; some shares are at an 80% discount, some of them are selling at 10% discounts. A lot depends on what that last round looked like. If you raised at 100x revenue in 2021 from SoftBank at a really competitively-led round, we’re seeing discounts that are wider than 40% compared with companies that raised capital in the first quarter or two of this year at a more ‘relatable’ valuation, where you might see a more modest discount.

I wouldn’t say that we’ve seen a bounce back on valuations. I will say that the acceleration downwards is slowing down, so we’re not seeing shares go from 40% to 60% immediately. And so my guess is if more trades start to happen at this 40% range, particularly involving large institutions and known institutions, it may indicate that we’re either going to sit at this floor or we’re going to start to bounce back. [But] a lot of it remains dependent on performance in the public markets. If we continue to see the Nasdaq trade down another 5% to 10% and the high-beta names in the public markets trade down 20% or 30%, you’ll see [share value] in the secondary markets continue to go down.

How much has EquityZen raised from VCs over the years?

A little under $7 million. We’re a very boring company as far as venture backing goes. We last raised money in February 2017. We’ve really relied on the business model and profitability of the business to reinvest and grow.

I would say it’s probably the hardest thing we’ve had to do here at EquityZen by far, letting go of some really, really good people [last month]. But the benefit of being a company that hasn’t raised too much outside funding is that it was a decision we made when we wanted to make it. It wasn’t something that a board told us we had to do before by XYZ date.

A rival of yours, Forge Global, went public back in March through a SPAC and its timing didn’t help but its shares are trading at $1.33. Its market cap is just $230 million, which is less than the $238 million that investors had poured into the company when it was still private. How does that impact how you’re thinking about next steps?

We’re still very much in the early innings. We want to be able to continue to bring private markets to the public. And if that means that it’s doing it as a public company, that’s fine. If it means doing it as a privately held company, that’s also fine. If that means doing it as part of a larger financial services business, that’s also okay, so long as we can continue to work on it. We’ve got about 250,000 accredited investors on the platform. We’ve transacted in a little over 400 private technology companies to date. I really do think we’re just starting to scratch the surface.

On the secondary market, shares are discounted 40% on average, says industry pro by Connie Loizos originally published on TechCrunch

Menstrual cups (sometimes known as “moon cups”) have been around since the 1930s, and it’s been a bloody long period since we’ve seen much innovation in that category. Emm just closed a $1 million round of funding to add some smarts. The company adds an applicator and an app tracking feature to help keep an eye on things.

We spoke with the company’s founder to learn more about where she sees the market going in the near future.

“What drives me is the firm belief that information about your body can make a life-altering difference to your health outcomes. That’s why what we’re doing at Emm is important — we’re addressing health with our sensing platform and we’re also addressing quality of life with product performance. And that combination of innovation in technology with design is magic,” says Emm founder & CEO Jenny Button in an interview with TechCrunch. “My vision for the next generation of consumer health technology makes me the perfect person to run Emm, and it means that my team (who help bring that vision to life) and I are working to deliver meaningful impact in what’s one of the most important sectors in the world.”

The £891,000 round ($1.1 million) came from a mix of angel groups and the government org Innovate UK. The goal of the round is to hire the core team, and work toward broader market validation. The company is currently in a stage of beta testing, and is planning a product launch next year. The company has been working on developing and testing its product for the past two years.

Emm isn’t the first smart menstrual cup on the market. Most notably, Looncup took a stab at the market with a Kickstarter project back in 2015, which raised $160,000 or so, and was scheduled to ship its products in 2016. From the backer comments, it appears that the product never shipped, and the last update from the Looncup team was that the company received FDA approval in 2019. Looncup did not reply to a request for a comment for this article.

“Our ultimate goal is to progress individual and collective health through new and unique bio-data. And if it all goes to plan, consumers will have access to tools that don’t currently exist that will provide unique insight into how their body is working and their state of health,” explains Button. “This means they’ll be able to manage their healthcare better thanks to this next generation of wearable technology.”

Like many businesses, the company had a number of wobbles over the past couple of years, largely due to supply chain disruptions.

“My favorite thing about working on Emm is working with a brilliant team of designers, researchers, engineers and scientists who are equally passionate and committed to improving health outcomes,” Button says.

To da moon: Emm aims to innovate on the menstrual cup by Haje Jan Kamps originally published on TechCrunch

Boundary Layer, which was gunning for local air freight, and announced a slew of launch partners earlier this year, today announced a shift in strategy, with some high-performance foiling personal watercraft. Think low-flying SeaDoo, and you’ve got the right idea.

The new product is called Valo, and will be able to carry a couple of passengers. Powered by a 108 hp electric motor, the craft can foil two feet above the surface of the water, at a top speed of 58 mph (50 knots), which will make it the world’s fastest production foiling craft. Fingers crossed you don’t come off the craft at those speeds; that’d be profoundly uncomfortable.

“We advise that you don’t fall off at 50 knots,” Ed Kearney, founder and CEO of the company behind Valo, Boundary Layer Technologies, said drily.

The company says the hydrofoils will be fully retractable, which would enable trailering on a conventional boat trailer.

“Valo will be a complete revolution to personal watercraft. The first Jet Ski was on the market 50 years ago this year, and it’s time for a major upgrade,” Kearney said. “It will be fast, agile and tremendously exhilarating, all while being near silent and leaving zero wake. It will be like flying a stunt plane but on water. We see this as a completely new form of water-based mobility.”

It’s a curious pivot for the company that was previously focusing on commercial foiling passenger ferries. Moving from light, high-speed shipping to a leisure craft can’t have been an easy choice for the Y Combinator-backed company, which announced it had raised a $4.8 million round from Lower Carbon Capital,Fifty YearsandSoma Capital. At the time, the company claimed it had $90 million worth of preorders from ferry operators for their 220-seat electric passenger vessels.

“The other projects are on hold. We see a huge opportunity for electric foiling, ferries and container ships to replace fright. For us, the question is ‘where do we start?’,” Kearney told us in an interview. “We had letters of intent for those vehicles, mostly passenger ferries. The customers were looking for the next step of technical progress. But it has been a very challenging time for them, between the global environment in terms of oil prices and geopolitical turmoil. They will be waiting keenly for us to get to the next stage, where they can have their vessels.”

So the original set of customers will have to be a bit patient for now — but why the pivot to personal watercraft?

“I’ve been on a lot of foiling boats, but nothing quite lives up to the full potential that foiling should be bringing to a watercraft. It should be going like downhill skiing, carving through deep powder, or on a road bike going down a hill leaning into corners. This is the kind of thing that we can bring to the experience,” Kearney laughs. “That doesn’t exist today. And I really want to rip from the San Francisco Marina to Sausalito in eight minutes.”

The company’s founder says the pivot is more about a shift of focus.

“We simply shifted from ‘big first’, to ‘fast first’,” says Kearney. “What we love about Valo is how fast we can get to market. We are bringing all the technology we were developing for massive container ships and ferries and using it to deliver one hell of a recreational product.”

The company claims that the design and build of the first prototype of this craft is “almost complete” and that the company expects to start offering the first customer demos in Q1 2023. The company will be building a small number of limited edition “Founders Edition” craft by mid 2023, before releasing the production vehicle in 2024, with an expected price tag of $59,000.

Boundary Layer pivots from container ships to hydrofoiling personal watercraft by Haje Jan Kamps originally published on TechCrunch

Tesla is offering Model 3 and Model Y buyers in the U.S. a $3,750 credit if they have their vehicle delivered in December 2022, according to an update on the company’s inventory page and several posts on Reddit.

Tesla didn’t share the reason for such generosity or publish the news widely — customers received the offer through messages from their dealers. It’s possible the automaker wants owners to take their deliveries before the year is out so it can boost its fourth-quarter sales numbers, which might be dwindling as customers push their delivery times out to 2023 in order to be eligible for the electric vehicle tax credit.

Tesla’s cars haven’t been eligible for an EV tax credit for some time — since the automaker reached the previous cap of 200,000 vehicles sold. But with President Biden’s Inflation Reduction Act (IRA), that cap will be waived by January 1.

Under the new legislation, zero emissions vehicles will be eligible for up to $7,500 in tax credits if automakers can show that their battery components were made or assembled in North America and that a certain percentage of battery critical materials were extracted or processed in countries with which the U.S. has a free trade agreement. If an automaker can only show half, then they’ll only get half of the rebate, which will probably be the case for the next few years, considering most critical materials are still sourced from China.

Which brings us back to the $3,750 discount — the same amount of money a Tesla buyer is expected get back as a rebate next year. But instead of waiting for tax season, Tesla’s offer allows customers to reap the benefits now.

Tesla’s discount also doesn’t discriminate based on income or vehicle manufacturer suggested retail price (MSRP), according to messages received by Tesla customers. Under the IRA’s tax incentive, single tax filers are eligible if their income is below $150,000; heads of households if income is below $225,000; and joint filers’ if income is below $300,000. Additionally new electric cars and SUVs that cost more than $55,000 and $80,000, respectively, don’t qualify for the tax credit.

The Model 3, a compact car, starts at around $47,000 but can easily go over $66,000 depending on model, trim and year. And the 2022 Tesla Model Y starts at $64,990, but a fully loaded performance model can go for more than $80,000.

It’s not common for Tesla to provide discounts — in fact, it’s more like the automaker to increase its price. So the upcoming tax credit might not be the only reason Tesla is dangling discounts. A new report from S&P Global Mobility found that Tesla’s market dominance in the U.S. is waning. The company still dominates the EV sector with its 65% market share, but that’s down from 79% in 2020 and might drop below 20% by 2025. The loss of market share comes as other automakers roll out more affordable EVs.

Tesla offers $3,750 discount for Model 3, Model Y deliveries in December by Rebecca Bellan originally published on TechCrunch

Monarch Tractor this morning announced the delivery of its first MK-V unit. The “smart tractor” is electric and what the Bay Area-based company refers to as “driver optional” (terms like “autonomous” and “self-driving” come with their own unique baggage). We’ll just refer to it as a “robot tractor” from here on.

The system was unveiled a bit under two or so years ago. The timing was certainly right. The average age of a U.S. farmer is around 55, and hiring people to help out has become increasingly difficult. Agtech robotics is absolutely a category to be watching closely over the next few years, even amid rough economic headwinds for venture capital.

The first unit is going to wine producer Constellation Brands, which bought a half dozen of the tractors. Additional shipments “to family farms and other large multinational corporations” are also on their way.

“This is a momentous day for Monarch Tractor that has been years in the making,” Monarch co-founder and CEO Praveen Penmetsa said in a release. “Our team has worked tirelessly and relentlessly with a mission to make farming more profitable and sustainable. We’re proud to see our commitment to technological innovation and sustainable solutions culminate in celebrating our first tractor off the production line, which, I might add, is powered by renewable energy.”

The system is powered, in part, by Nvidia’s Jetson platform. That helps manage both driver assistance and the aforementioned “driver-optional modes.” There are several options for that implementation, including a “show mode,” wherein the system follows a worker. That’s managed, in part, by the inclusion of 360-degree camera systems for monitoring its surroundings and implementing various safety features.

Those imaging systems also play a role in another big piece of this. Data collection is arguably the most important role these sorts of systems play. Here that means the ability to monitor larger trends in crop health, yield and growth. These sorts of things are extremely important to farmers when it comes to managing current and future crops.

Monarch has thus far raised $100 million, including a $61 million round last year. It is set to begin manufacturing its systems in Ohio starting next year.

Monarch delivers its first robot tractor by Brian Heater originally published on TechCrunch

Motional, the Aptiv-Hyundai joint venture developing autonomous vehicle technology, cut its workforce this week — the latest example of layoffs in the AV industry and more broadly in the tech sector.

Employees were told of the layoffs Wednesday, according to sources who asked not to be named because they are not authorized to speak for the company. Motional, which employs more than 1,500 people globally, confirmed the layoffs.

Motional did not confirm the number of employees affected. Sources said dozens of employees were laid off with cuts happening across its operations. Motional has offices in Boston, Las Vegas, Pittsburgh, South Korea, Singapore and several cities in California, including Milpitas and Santa Monica.

“Motional recently announced steps to reallocate resources to areas of the company that will help ensure long-term commercial success,” a company spokesperson said in an emailed statement. “As the organization evolves, we will continue to hire the critical talent needed to develop our technology and meet our commitments to our partners. With our recent partnership announcements and the achievement of technology milestones, we are confident this shift will position us for a strong future.”

The company said it is providing severance packages and outplacement support to help in their transition.

Layoffs have become commonplace in the tech industry this year. The autonomous vehicle sector has been especially hard hit as the high development costs and lack of revenue — never mind profits — has left companies with dwindling cash. Investors, once ready to plow money into this frontier tech, are now looking for safer, nearer term bets.

In November, AV delivery startup Nuro laid off about 300 people, or 20% of its workforce, in an effort to preserve cash amid a stormy economic outlook. Argo AI, the AV startup backed by Ford and Volkswagen, also laid off employees earlier this summer. Several months later, Argo abruptly shut down after its two backers declined to invest more capital into the enterprise and shift its resources to advanced driver assistance systems.

While Motional didn’t provide a reason for its reallocation of resources, other AV companies have cited uncertain economic conditions as well as over-hiring in 2021 as a primary reason. Even AV companies that have yet to lay off employees are feeling pressure to cut costs. Aurora Innovation, a startup-turn-SPAC, presented a swath of cost-cutting and cash-generating options to its board earlier this year, ranging from a hiring freeze and spinning out assets to a small capital raise, going private and even selling itself to high-profile tech companies Apple and Microsoft.

Motional, like many of its remaining AV competitors, has focused its commercialization efforts on launching robotaxi services. The company sees partnerships as a key piece of its go-to-market strategy and has agreements with Lyft, Uber and Via.

Hyundai-backed AV startup Motional cuts workforce by Kirsten Korosec originally published on TechCrunch

To get a roundup of TechCrunch’s biggest and most important stories delivered to your inbox every day at 3 p.m. PDT, subscribe here.

There is officially only 8.3% left of the year. You know what that means! Holidays, cheer, and daydreaming about what shenanigans we’ll get ourselves into in 2024. Yes, 2024, because clearly we’re just going to skip past all of 2023. We’re pretty excited for 2025 already, tbh. — Christine and Haje

The TechCrunch Top 3

Celebrity strong arm: Plant-based food brand Huel’s newest capital raise comes with a side of Idris Elba, Christine writes. The actor and his wife are among a handful of new investors backing U.K.-based Huel, which offers protein powder, snack bars and hot lunch products. Oh, and Huel is now valued at $560 million. Not too shabby.
Who’s calling?: Why, it’s smartphone re-commerce startup Badili, which raised $2.1 million in pre-seed funding to work with individuals and dealers to repurpose phones as demand soars across Africa for more affordable phones. Annie has more.
Returning to its roots: Zoe made a name for itself during the global pandemic when it launched a COVID-tracking app. Now it has shifted back to its roots as a self-reporting tracker for personal nutrition and has a healthy $30 million chunk of change (and a $303 million valuation) under its belt to onboard over 250,000 people who have been patiently waiting for this moment, Ingrid reports.

Startups and VC

FTX and Sam Bankman-Fried have had a lot of column inches on Ye Olde Teche Crunche over the past couple of days. Here’s what you need to get you up to speed: Natasha M covered that SBF claims massive ignorance on obvious conflicts in FTX downfall. Amanda adds that he says journalists are good, actually, while Connie wonders if his interview appearance was a bit of a performance.

Web3 developer platform Fleek has raised $25 million in Series A funding led by Polychain Capital, the company told Jacquelyn. The startup is aiming to build an interface and protocol layer “to make the base layer of web[3] services,” like storage, hosting and billing, accessible to anyone.

And we have five more for you:

Please hold while we pause for repairs: Sarah writes that Twitter alternative Hive shuts down its app to fix critical security issues.
Your reward is…a slice of the stock market!: X1 gets a 50% valuation boostas the company aims to give consumers a way to buy stocks via credit card reward points, Mary Ann reports.
Tipping the scales: Brian writes that Lumen raises $62 million for its handheld weight-loss hardware.
Booting up: Haje took a closer look at Silicon Valley Bank–backed StartupOS, which launched what it hopes will be the operating system for early-stage startups.
A growth industry: Indian agritech company DeHaat tops $700 million valuation as it raises a $60 million round, Manish writes.

Proptech in Review: 3 investors explain why they’re bullish on tech that makes buildings greener

Investors who work at the intersection of climate tech and proptech seek out potentially profitable startups that can reduce emissions and enhance the built environment.

It’s a high-stakes balancing act with significant risk, but considering the upside for category winners and the health of the planet, “the potential market is enormous,” reports Tim De Chant. For his second proptech investor survey in a three-part series, he interviewed:

Jake Fingert, managing partner, and Lionel Foster, investor, Camber Creek
Anja Rath, managing partner, PropTech1 Ventures
Othmane Zrikem, chief data officer, A/O Proptech

Three more from the TC+ team:

CVC meets accelerators: Alex explores how Up.Labs threads the needle between corporate venture capital and accelerators.
Ads to the rescue: Alex Song shares 5 methods for leveraging digital advertising during a downturn.
Where’s your plans, friends? Haje is back with his weekly Pitch Deck Teardown, this time putting Hour One’s $20M Series A deck in the X-ray machine.

TechCrunch+ is our membership program that helps founders and startup teams get ahead of the pack. You can sign up here. Use code “DC” for a 15% discount on an annual subscription!

Big Tech Inc.

Before we step into the news, we just want to offer the AWS re:Invent 2022 link once again because Kyle, Ron and Frederic added more stories today.

Okay, here we go. Yesterday, Bret Taylor shocked us by stepping down as co-chair and CEO of Salesforce to “return to his entrepreneurial roots,” Ron reports. After ruminating on it overnight, Ron also provided more insight on Taylor’s move and what might be next for the CRM giant (TC+ subscription needed).

If you rush over to Google every day to see the new Doodle, the latest lets you create your own mini arcade game in honor of the late video game pioneer Jerry Lawson, who would have been 82 today. Aisha has more.

Meanwhile, Brian wrote a story yesterday reporting that “San Francisco police can now use robots to kill.” Today he goes more in-depth in his Actuator column. Enjoy!

And we have five more for you:

Today’s Twitter update: We know it’s been mere seconds since we last read what was on Elon Musk’s mind, but Taylor reports that your Twitter feed may soon be controlled by a higher force, while Natasha L writes that Musk may lose even more sleep in order to get Twitter in compliance with European Union rules.
This might be one club you want in on: Netflix is reportedly letting more subscribers preview its films and TV shows and provide their feedback, Lauren writes.
This might be one inbox you don’t check: If you’re upset that your LinkedIn inbox has become a place where spam and scams live, Ingrid has something you might like: focused inbox and messaging safety tools to rid you of some of that.
This might be one too many: Password manager LastPass has more bad news: it says it was breached — again, Zack writes.
Gadgets and gizmos aplenty: Brian has curated a list of go-to gifts for frequent flyers.

Daily Crunch: SBF says he’s ‘had a bad month,’ but is he really giving us the full story? by Christine Hall originally published on TechCrunch

While artificial intelligence long ago surpassed human capability in chess, and more recently Go — and let us not forget Doom — other more complex board games still present a challenge to computer systems. Until very recently, Stratego and Diplomacy were two of those games, but now AI has become table-flipping good at the former and passably human at the latter.

On the surface, you might think that it’s just because these games require a certain level of long-term planning and strategy. But so do Go and chess, just in a different way.

The crucial difference is actually that Stratego and Diplomacy are games of strategy based on imperfect information. In chess and Go, you can see every piece on the board. Stratego hides the identity of pieces until they are encountered by another piece, and Diplomacy is largely about establishing agreements, alliances and, of course, vendettas that are kept secret but are core to the gameplay. No honest chess game will involve a third party swooping in to protect your opponent’s bishop with a blue rook.

Both games require not raw calculation of paths to victory, but softer skills like guessing what the opponent is thinking, and what they think the computer is thinking, and make moves that accommodate and hopefully upset those assumptions. In other words, it has to bluff and convince another player of something, not just overpower it with the best possible moves.

The Stratego-playing model, from DeepMind, is named DeepNash, after the famous equilibrium. It is focused less on clever moves and more on play that can’t be exploited or predicted. In some cases this can be bold, like one game the team watched against a human player where the AI sacrificed several high-level pieces, leaving it at a material disadvantage — but it was all a calculated risk to bring out the other player’s big guns, so it could strategize around those. (It won.)

DeepNash is good enough that it beat other Stratego systems almost every time, and 84% of the time versus experienced humans. Because the algorithms that work well in Go and chess don’t work well here, they invented a new algorithmic method called Regularised Nash Dynamics — but you’ll have to read the paper if you want to understand it any more deeply than that. In the meantime, here’s an annotated game:

On the Diplomacy side, we have an AI named Cicero (ah, hubris!) from Meta and CSAIL that manages to play the game at a human level — and if that sounds like damning with faint praise, remember Diplomacy is difficult for most humans to play at a human level. The level of scheming, backstabbing, false promises and general Machiavellian antics that people get up to in the game are such that it is banned from many friendly gaming groups. Is a computer really capable of that level of shenanigans?

Seems so, and the advances that make it possible are interesting. After all, the interesting part of Diplomacy isn’t the world map and pieces, which are fairly straightforward to read and evaluate, but the potential for schemes latent in those arrangements. Is Venice being threatened on two fronts, or is it luring the western front into an envelopment through a long contemplated volte-face?

Not only that, but in order to participate in the scheming, one must speak (or chat, online) to other players and convince them of your sincerity and intent. This takes more than CPU cycles!

Here’s how Cicero works:

Using the board state and current dialogue, make an initial prediction of what everyone will do.
Refine that prediction using planning and then use those predictions to form an intent for itself and its partner.
Generate several candidate messages based on the board state, dialogue and its intents.
Filter the candidate message to reduce nonsense, maximize value and ensure consistency with our intents.

Then, plea your case and hope the other player isn’t planning your demise.

When set loose on webDiplomacy.net, Cicero played quite well against its opponents, placing 2nd out of 19 in a league and generally outscoring others.

It’s still very much a work in progress — it can lose track of what it’s said to others, or make other blunders humans probably wouldn’t — but it’s pretty remarkable that it can be competitive at all.

Now AI can outmaneuver you at both Stratego and Diplomacy by Devin Coldewey originally published on TechCrunch

The technological advances happening within the space industry are nothing short of dramatic — from propulsion systems, heavy-lift rockets and increased payload capabilities to building commercial space stations to fill the gap when the International Space Station ceases operations in 2030.

These are thrilling times, and TC Sessions: Space 2022, which takes place on December 6 in Los Angeles, is where you’ll meet the leading founders, investors and makers building the future of space exploration today.

Countdown to launch: Buy a pass now and be in the room to learn, engage, connect and network your way to a stronger business.

Take a deep dive through the impressive agenda to plan your schedule, and be sure to save time to meet and greet the early-stage startups exhibiting equally impressive technology, products and possibilities on the show floor.

Meet the startups exhibiting at the show:

Astrum Drive Technologies
At Astrum Drive Technologies we merge science and technology.

Innovations Health Systems, LLC
Our approach is critical to long-term human health & fitness in MGE

JELLYSPACE
JELLYSPACE removes the entry barriers into the space industry by allowing companies, Startups and organizations to share and access Intellectual Property in safe and secure manner.

Morpheus Space
Morpheus Space is redefining satellite propulsion, serving as the support for those innovators chasing the exciting dream of space exploration.

Nooroot Inc.
Nooroot is a deep-tech company which offers unique data analysis and visualization capabilities inspired by theoretical physics and pure mathematic principles such as the Superstring theory. Our platform technology enables us to unify, analyze, and visualize streaming aeronautical data as well as terrestrial data from any source/sensor in real-time transparently and sustainably.

Plasmos Inc
Reusable last mile delivery for satellites with a unique propulsion system

TRL11, Inc
We make space become more than just mature (TRL9), we make it become mainstream, (TRL11).

Vermeer
Vermeer builds vision based Mixed Reality & AI Enabled Visualization and Command and Control navigation systems for GPS denied and degraded environments. Our solution works on ground vehicles, mobile devices, aircraft, and sUAS.

We’re excited to host some of the most influential leaders, investors and decision-makers in the space industry. You’ll hear from Jory Bell, general partner at Playground Global; Frank Calvelli, assistant secretary of the Air Force for Space Acquisitions; Steve Jurczyk, co-founder and CEO of Quantum Space; Melanie Stricklan, co-founder and CEO at Slingshot Aerospace; and many more.

TC Sessions: Space takes place on December 6 in Los Angeles. Buy your pass today, and then join us to see and learn about the latest tech from established space companies, rising-star founders, and then network for opportunities to build a stronger startup.

Is your company interested in sponsoring or exhibiting at TC Sessions: Space? Contact our sponsorship sales team byfilling out this form.

Meet the early-stage startups exhibiting at TC Sessions: Space by Lauren Simonds originally published on TechCrunch

Released two years ago, OpenAI’s remarkably capable, if flawed, GPT-3 was perhaps the first to demonstrate that AI can write convincingly — if not perfectly — like a human. The successor to GPT-3, most likely called GPT-4, is expected to be unveiled in the near future, perhaps as soon as 2023. But in the meantime, OpenAI has quietly rolled out a series of AI models based on “GPT-3.5,” a previously-unannounced, improved version of GPT-3.

GPT-3.5 broke cover on Wednesday with ChatGPT, a fine-tuned version of GPT-3.5 that’s essentially a general-purpose chatbot. Debuted in a public demo yesterday afternoon, ChatGPT can engage with a range of topics, including programming, TV scripts and scientific concepts.

According to OpenAI, GPT-3.5 was trained on a blend of text and code published prior to Q4 2021. Like GPT-3 and other text-generating AI, GPT-3.5 learned the relationships between sentences, words and parts of words by ingesting huge amounts of content from the web, including hundreds of thousands of Wikipedia entries, social media posts and news articles.

Rather than release the fully trained GPT-3.5, OpenAI used it to create several systems fine-tuned for specific tasks — each available through the OpenAI API. One — text-davinci-003 — can handle more complex instructions than models built on GPT-3, according to the lab, and is measurably better at both long-form and “high-quality” writing.

According to OpenAI data scientist Jan Leike, text-davinci-003 is similar but not identical to InstructGPT, a family of GPT-3-based models released by OpenAI earlier this year that are less likely to generate problematic (e.g., toxic and highly biased) text while more closely aligning with a user’s intent. Text-davinci-003 — and by extension GPT-3.5 — “scores higher on human preference ratings” while suffering from “less severe” limitations, Leike said in a tweet.

That appears to be the case anecdotally. Data scientists at Pepper Content, a content marketing platform, report that text-davinci-003 “performs better in understanding the ‘context’ behind a request and then using that to produce better content” while “hallucinating” less than GPT-3-based models. (Where it concerns text-generating AI, hallucination refers to an AI writing inconsistent, factually incorrect statements.)

In a test on OpenAI’s Playground website, which provides a UI frontend for the models, the Pepper Content team fed several prompts to text-davinci-003 and a model based on GPT-3 (text-davinci-002). Given “What is the philosophy behind WeWork?,” the GPT-3.5-based text-davinci-003 generated this:

WeWork is a global workspace provider that believes people are the most important asset in any organization. The philosophy of WeWork is to create a collaborative environment that enables people to work together in a flexible and efficient way. The company encourages collaboration and productivity, while providing a comfortable and inspiring space.

WeWork is also committed to being a socially responsible organization, by finding ways to reduce its environmental impact, by providing meaningful work experiences, and by promoting diversity and inclusion. WeWork also strives to create meaningful experiences for its members, through its unique community-based programming, events and activities. The company believes that when people work together in an inspiring and collaborative environment, they can achieve more and create meaningful change.

It’s not perfect — note the excess commas and repetitiveness. But the copy’s certainly more engaging than what the GPT-3-based text-davinci-002 produced:

At WeWork, we believe in openness, collaboration, and community. We have reimagined what a workspace can be by bringing together a global community of creators, entrepreneurs, and startups — anyone looking to build something meaningful and transform the world. We are proud to be your neighbour, coworker, and friend.

GPT-3.5 is also better at generating blog posts, it seems. Here’s what the Pepper Content team got when they prompted text-davinci-003 to write a post about picking a sofa:

If you are looking to give your green living room a fresh look, adding the the best sofa for the space is a great way to start. However, there is more to choosing the perfect sofa than just finding a piece that looks good. You want to find a piece that is comfortable and fits your green living room décor.

In this blog, we will discuss how to pick sofa far your green living room. We will look at the different types of sofas available, the design factors to consider when making your selection, and how to choose a sofa that fits your style and budget. We will also provide tips on how to maintain and care for your new sofa. By the end of this blog, you will have all the information you need to pick the perfect for your green living room.

Again, it isn’t foolproof. GPT-3.5 oddly added the bit about a “green living room.” But also again, GPT-3 is more basic and less grammatical in its generation:

Sofa is one of the most basic requirements in a living room. It’s not just a piece of furniture but an important part of the décor of your living room. So, what should be the criteria while picking a sofa? If you are wondering about this then stay with me as I discuss the different aspects of the sofa would help you in picking the best one for yourself.

Experiments beyond Pepper Content’s suggest that GPT-3.5 tends to be much more sophisticated and thorough in its responses than GPT-3.

For example, when YouTube channel All About AI prompted text-davinci-003 to write a history about AI, the model’s output mentioned key luminaries in the field, including Alan Turing and Arthur Samuelson, while text-davinci-002”s did not. All About AI also found that text-davinci-003 tends to have a more nuanced understanding of instructions, for instance providing details such as a title, description, outline, introduction and recap when asked to create a video script.

A hallmark feature of text-davinci-003/GPT-3.5’s generations is wordiness, as it turns out. In an analysis, scientists at startup Scale AI found text-davinci-003/GPT-3.5 generates outputs roughly 65% longer than text-davinci-002/GPT-3 under identical prompts.

Perhaps less useful but nonetheless entertaining, text-davinci-003/GPT-3.5 is better at composing songs, limericks and rhyming poetry than its predecessor. Ars Technica reports that commenters on Y Combinator’s Hacker News forum used text-davinci-003 to write a poem explaining Albert Einstein’s theory of relativity and then re-write the poem in the style of John Keats. See:

If you want to understand Einstein’s thought
It’s not that hard if you give it a shot
General Relativity is the name of the game
Where space and time cannot remain the same
Mass affects the curvature of space
Which affects the flow of time’s race
An object’s motion will be affected
By the distortion that is detected
The closer you are to a large mass
The slower time will seem to pass
The farther away you may be
Time will speed up for you to see

The Scale AI team even found that Text-davinci-003/GPT-3.5 has a notion of meters like iambic pentameter. See:

O gentle steeds, that bear me swift and

sure

Through fields of green and pathways so

obscure,

My heart doth swell with pride to be with

you

As on we ride the world a-fresh to view

The wind doth whistle through our hair so

free

And stirs a passion deep inside of me.

My soul doth lift, my spirits soar on high,

To ride with you, my truest friend, am I

Your strength and grace, your courage and

your fire,

Inspire us both to go beyond our sire.

No earthly bonds can hold us, only fate,

To gallop on, our wond’rous course create

Relatedly, text-davinci-003/GPT-3.5 is wittier — at least subjectively. Asking text-davinci-002/GPT-3 to “tell a joke” usually yields this:

Why did the chicken cross the road? To get to the other side.

Text-davinci-003/GPT-3.5 has cleverer responses:

Q: What did the fish say when it hit the wall? A: Dam!

Q: What did one ocean say to the other ocean? A: Nothing, they just waved.

Scale AI had it explain Python code as Eminem, a feat which text-davinci-002/GPT-3 couldn’t accomplish:

Yo, so I’m loopin’ through this list

With each item that I find

I’m gonna print out every letter in each one

of them

Dog, Cat, Banana, Apple, I’m gonna get’em

all with this rhyme

So why is GPT-3.5 so much better than GPT-3 in these particular areas? We can’t know the exact answer without additional details from OpenAI, which aren’t forthcoming; an OpenAI spokesperson declined our request for comment. But it’s safe to assume that GPT-3.5’s training approach had something to do with it. Like InstructGPT, GPT-3.5 was trained with the help of human trainers who ranked and rated the way early versions of the model responded to prompts. This information was then fed back into the system, which tuned its answers to match the trainers’ preferences.

Of course, it doesn’t make GPT-3.5 immune to the pitfalls to which all language models eventually succumb. Because GPT-3.5 merely relies on statistical regularities in its training data rather than a human-like understanding of the world, it’s still prone to, in Leike’s words, “mak[ing] stuff up a bunch.” It also has limited knowledge of the world after 2021 because the training data is more sparse after that year. And its safeguards against toxic output can be straightforwardly circumvented.

Still, GPT-3.5 and its derivative models demonstrate that GPT-4 — whenever it arrives — won’t necessarily need a huge number of parameters to best the most capable text-generating systems today. (Parameters are the parts of the model learned from historical training data and essentially define the skill of the model on a problem.) While some have predicted that GPT-4 will contain over 100 trillion parameters — nearly 600 times as many as GPT-3 — others argue that emerging techniques in language processing, like those seen in GPT-3.5 and InstructGPT, will make that enormous jump unnecessary.

One of those techniques could involve browsing the web for greater context, a la Meta’s ill-fated BlenderBot 3.0 chatbot. John Shulman, a research scientist and co-founder of OpenAI, told MIT Tech Review that OpenAI is continuing work on a language model it announced late last year, WebGPT, that can go and look up information on the web (via Bing) and give sources for its answers. At least one Twitter user appears to have found evidence of the feature undergoing testing for ChatGPT.

OpenAI has another reason to pursue lower-parameter models as it continues to evolve GPT-3: huge costs. A 2020studyfromAI21 Labs pegged the expenses for developing a text-generating model with only 1.5 billion parameters at as much as $1.6 million. OpenAI has raised over $1 billion to date from Microsoft and other backers, and it’s reportedly in talks to raise more. But all investors, no matter how big, expect to see returns eventually.

While anticipation builds for GPT-4, OpenAI quietly releases GPT-3.5 by Kyle Wiggers originally published on TechCrunch