Connect with us

Uncategorized

This is how we lost control of our faces

Published

on

In 1964, mathematician and computer scientist Woodrow Bledsoe first attempted the task of matching suspects’ faces to mugshots. He measured out the distances between different facial features in printed photographs and fed them into a computer program. His rudimentary successes would set off decades of research into teaching machines to recognize human faces.

Now a new study shows just how much this enterprise has eroded our privacy. It hasn’t just fueled an increasingly powerful tool of surveillance. The latest generation of deep-learning-based facial recognition has completely disrupted our norms of consent.

Deborah Raji, a fellow at nonprofit Mozilla, and Genevieve Fried, who advises members of the US Congress on algorithmic accountability, examined over 130 facial-recognition data sets compiled over 43 years. They found that researchers, driven by the exploding data requirements of deep learning, gradually abandoned asking for people’s consent. This has led more and more of people’s personal photos to be incorporated into systems of surveillance without their knowledge.

It has also led to far messier data sets: they may unintentionally include photos of minors, use racist and sexist labels, or have inconsistent quality and lighting. The trend could help explain the growing number of cases in which facial-recognition systems have failed with troubling consequences, such as the false arrests of two Black men in the Detroit area last year.

People were extremely cautious about collecting, documenting, and verifying face data in the early days, says Raji. “Now we don’t care anymore. All of that has been abandoned,” she says. “You just can’t keep track of a million faces. After a certain point, you can’t even pretend that you have control.”

A history of facial-recognition data

The researchers identified four major eras of facial recognition, each driven by an increasing desire to improve the technology. The first phase, which ran until the 1990s, was largely characterized by manually intensive and computationally slow methods.

But then, spurred by the realization that facial recognition could track and identify individuals more effectively than fingerprints, the US Department of Defense pumped $6.5 million into creating the first large-scale face data set. Over 15 photography sessions in three years, the project captured 14,126 images of 1,199 individuals. The Face Recognition Technology (FERET) database was released in 1996.

The following decade saw an uptick in academic and commercial facial-recognition research, and many more data sets were created. The vast majority were sourced through photo shoots like FERET’s and had full participant consent. Many also included meticulous metadata, Raji says, such as the age and ethnicity of subjects, or illumination information. But these early systems struggled in real-world settings, which drove researchers to seek larger and more diverse data sets.

In 2007, the release of the Labeled Faces in the Wild (LFW) data set opened the floodgates to data collection through web search. Researchers began downloading images directly from Google, Flickr, and Yahoo without concern for consent. LFW also relaxed standards around the inclusion of minors, using photos found with search terms like “baby,” “juvenile,” and “teen” to increase diversity. This process made it possible to create significantly larger data sets in a short time, but facial recognition still faced many of the same challenges as before. This pushed researchers to seek yet more methods and data to overcome the technology’s poor performance.

Then, in 2014, Facebook used its user photos to train a deep-learning model called DeepFace. While the company never released the data set, the system’s superhuman performance elevated deep learning to the de facto method for analyzing faces. This is when manual verification and labeling became nearly impossible as data sets grew to tens of millions of photos, says Raji. It’s also when really strange phenomena start appearing, like auto-generated labels that include offensive terminology.

The way the data sets were used began to change around this time, too. Instead of trying to match individuals, new models began focusing more on classification. “Instead of saying, ‘Is this a photo of Karen? Yes or no,’ it turned into ‘Let’s predict Karen’s internal personality, or her ethnicity,’ and boxing people into these categories,” Raji says.

Amba Kak, the global policy director at AI Now, who did not participate in the research, says the paper offers a stark picture of how the biometrics industry has evolved. Deep learning may have rescued the technology from some of its struggles, but “that technological advance also has come at a cost,” she says. “It’s thrown up all these issues that we now are quite familiar with: consent, extraction, IP issues, privacy.”

Harm that begets harm

Raji says her investigation into the data has made her gravely concerned about deep-learning-based facial recognition.

“It’s so much more dangerous,” she says. “The data requirement forces you to collect incredibly sensitive information about, at minimum, tens of thousands of people. It forces you to violate their privacy. That in itself is a basis of harm. And then we’re hoarding all this information that you can’t control to build something that likely will function in ways you can’t even predict. That’s really the nature of where we’re at.”

She hopes the paper will provoke researchers to reflect on the trade-off between the performance gains derived from deep learning and the loss of consent, meticulous data verification, and thorough documentation. “Was it worth abandoning all of these practices in order to do deep learning?” she says.

She urges those who want to continue building facial recognition to consider developing different techniques: “For us to really try to use this tool without hurting people will require re-envisioning everything we know about it.”

Continue Reading
Comments

Uncategorized

Proptech startup States Title, now Doma, going public via SPAC in $3B deal

Published

on

Real estate tech startup Doma, formerly known as States Title, announced Tuesday it will go public through a merger with SPAC Capitol Investment Corp. V in a deal valued at $3 billion, including debt.

SPACs, often called blank-check companies, are increasingly common. They exist as publicly traded entities in search of a private company to combine with, taking the private entity public without the hassle of an IPO.

When it floats later this year, Doma will trade on the New York Stock Exchange under the ticker symbol DOMA. The transaction is expected to provide up to $645 million in cash proceeds, including a fully committed PIPE of $300 million and up to $345 million of cash held in the trust account of Capitol Investment Corp. V. 

CEO Max Simkoff founded San Francisco-based Doma in September 2016 with the aim of creating a technology-driven solution for “closing mortgages instantly.” While it initially was founded to instantly underwrite title insurance, the company has expanded that same approach to handle “every aspect” of closing and escrow.

Doma has developed patented machine learning technology that it says reduces title processing time from five days to “as little as one minute” and cuts down the entire mortgage closing process “from a 50+ day ordeal to less than a week.” The startup has facilitated over 800,000 real estate closings for lenders such as Chase, Homepoint, Sierra Pacific Mortgage and others.

The name change is designed to more accurately reflect its intention to expand “well beyond” title into areas such as appraisals and home warranties.

Its goal with going public is to be able to “continue to invest in growth, market expansion and new products.”

Anchoring the PIPE include funds and accounts managed by BlackRock, Fidelity Management & Research Company LLC, SB Management (a subsidiary of SoftBank Group), Gores, Hedosophia, and Wells Capital. Existing Doma shareholder Lennar has also committed to the PIPE and Spencer Rascoff, co-founder and former CEO of Zillow Group, has committed a personal investment to the PIPE.

Up to approximately $510 million of cash proceeds are expected to be retained by Doma, and existing Doma shareholders will own no less than approximately 80 percent of the equity of the new combined company, subject to redemptions by the public stockholders of Capitol and payment of transaction expenses.

In mid-February, Doma announced it had closed on $150 million in debt financing from HSCM Bermuda, which had previously invested in the company. And last May, it announced a massive $123 million Series C round of funding at a valuation of $623 million.

Doma joins the growing number of proptech companies going the public route. On Monday, Compass, the real-estate brokerage startup backed by roughly $1.6 billion in venture funding, filed its S-1

In 2020, Social Capital Hedosophia II, the blank-check company associated with investor Chamath Palihapitiya, announced that it would merge with Opendoor, taking the private real estate startup public in the process.

Porch.com also went public in a SPAC deal in December. And, SoftBank-backed View, a Silicon Valley-based smart window company, will complete a recent SPAC merger to be publicly listed on the NASDAQ stock exchange on March 8. The company is expected to debut trading with a market value of $1.6 billion.

Continue Reading

Uncategorized

Fluid Truck, the Zipcar of commercial trucks, raises $63M to take on rental giants

Published

on

Fluid Truck has built an app-based platform that aims to take away the pain and cost of owning or leasing commercial vehicles, all while grabbing market share from established companies like Penske, Ryder and U-Haul. 

Now, it has the capital to help it get there. The Denver-based company said Tuesday it raised $63 million in a Series A funding round to expand its truck-sharing platform, which helps mid-mile and last-mile delivery companies remotely manage an on-demand rental fleet via web or mobile app. Private equity firm Bison Capital led the round, with participation from Ingka Investments (part of Ingka Group, the main Ikea retailer), Sumitomo Corporation of Americas and Fluid Vehicle Owners.  

The investment, its first external round, comes after rapid growth at the four-year-old company. Founder and CEO James Eberhard told TechCrunch that revenue increased 100x in the last two years. That type of growth sounds promising, but the company did not provide a baseline, so it’s hard to judge scale. 

With e-commerce expected to continue to rise at a global 9.5% compound annual growth rate from 2020 to 2025, the demand for accessible trucks for hire might see correlative growth. It’s no surprise that e-commerce is one of the industries Fluid Truck has targeted. 

Fluid Truck, which operates in 25 U.S. markets, operates like the car-sharing company Zipcar, with a commercial bent. Businesses such as moving and e-commerce delivery companies can use the platform to rent trucks. Fluid Truck’s pitch to businesses extends beyond the “you don’t need to buy or lease” argument. The platform also allows delivery companies to dispense with having a manager on staff who would manage, maintain and eventually sell the fleet. 

Businesses eager to outsource the purchasing and managing of their trucks can find fleets for hire in industrial parks and retail areas within Fluid’s service network. 

“You can hop on our platform, rent a truck and be in it in a matter of minutes, which really allows businesses to scale up and scale down,” said Eberhard. “We’re watching our user behavior go from a place where they used to own every vehicle they needed at a time to a place where they’re now grabbing spare capacity off Fluid.”

Eberhard hopes to see that type of supplementary use morph into an end state where companies don’t own a single truck and run solely on Fluid Truck’s platform. 

Fluid Truck argues that its tech stack, which is designed to smooth out the booking and renting process, gives it a competitive edge in a market dominated by the likes of U-Haul, Ryder and or other small depots. Eberhard said the process of going to a depot and waiting in line is slow and sloppy, whereas Fluid Truck’s app makes renting a van as easy as calling an Uber.

“We take all those complexities away and allow people to have a virtual fleet,” Eberhard told TechCrunch.

Fluid Truck’s fleet is made up of thousands — and soon to be tens of thousands — of cargo vans, pickup trucks, large box trucks and various other vehicles. The company also claims to have the largest medium-duty EV rental fleet in the United States, which it continues to expand as it works with OEMs to increase fleet capacity. Electric vehicles still make up less than 1% of its total portfolio due to the slower adoption of EVs on the commercial side. 

Eberhard wants Fluid to be a dominant force in the trucking industry. But Fluid Truck is not the only truck sharing app on the streets. Competitors GoShare and Bungii have similar offerings.

This sizable round could provide an advantage as it tries to become the household name in digital truck sharing. Perhaps, as importantly, the company has the attention and investment of Ikea. 

“This is another step in enabling Ikea retail to provide last mile delivery services to our customers, continue to improve on our customer promise, while also reducing our environmental footprint,” Krister Mattsson, managing director of Ingka Investments said in a statement, a comment that suggests a future partnership with Fluid Truck. 

With this latest capital round, Fluid’s goal is to (you guessed it) scale outwards, with a focus on expanding the team, adding dozens more markets in the U.S. and preparing to take Fluid into the EU and Canada. 

Fluid Truck will also be investing back into its own tech stack, which includes an internal proprietary telematics platform to predict and automate servicing and maintenance of the company’s fleet. 

Continue Reading

Uncategorized

Twitter Spaces arrives on Android ahead of Clubhouse

Published

on

Twitter announced today it’s opening up its live audio chat rooms, known as Twitter Spaces, to users on Android. Previously, the experience was only open to select users on iOS following the product’s private beta launch in late December 2020. The company says that Android users will only be able to join and talk in Spaces for the time being, but won’t yet be able to start their own.

That added functionality is expected to ship “soon,” Twitter says, without offering an exact timeframe.

The company has been working quickly to iterate on Twitter Spaces in the months since its beta debut, and has been fairly transparent about its roadmap.

Last month, the team developing Twitter Spaces hosted a Space where users were invited to offer feedback, ask questions, and learn about what Twitter had in the works for the product in both the near-term and further down the road. During this live chat, Twitter confirmed that Spaces would arrive on Android in March.

It also promised a fix to how it displays listeners, which has since rolled out.

Other Spaces features are being shared in public as they’re designed and prototyped, including things like titles and descriptions, scheduling options, support for co-hosts and moderators, guest lists, and more. Twitter has also updated the preview card that appears in the timeline and relabeled its “captions” feature to be more accurate, from an accessibility standpoint.

The time frame of some of its new developments  — like Android and scheduling options — were being promised in a matter of weeks, not months.

This fast pace has now led Twitter to beat its rival Clubhouse — the app currently leading the “social audio” market — to offer support for Android. Today, Clubhouse remains iOS-only in addition to being invite-only.

It’s also indicative of the resources Twitter is putting into this new product, which was first announced publicly just in November. Clearly, Twitter believes social audio is a market it needs to win.

The company also sees the broader potential for Spaces as being a key part of a larger creator platform now in the works. During its Investor Day last week, Twitter spoke of tying together its new products like Spaces, Newsletters along with a “Super Follow” paid subscription, for example.

It’s now also testing a Twitter “Shopping Card” that would allow users to tweets posts that link directly to product pages via a “Shop” button — a feature that would seem to fall under this new creator focus, as well.

Some Twitter users on Android had already found their way to Spaces before today’s announcement by way of the Twitter beta app on Google Play.

But now, a separate beta app won’t be required — when live Spaces are available, they’ll appear at the top of the Twitter timeline for Android users to join.

 

 

Continue Reading

Trending