4 questions to ask when evaluating AI prototypes for bias

It’s true there has been progress around data protection in the U.S. thanks to the passing of several laws, such as the California Consumer Privacy Act (CCPA), and nonbinding documents, such as the Blueprint for an AI Bill of Rights. Yet, there currently aren’t any standard regulations that dictate how technology companies should mitigate AI bias and discrimination.

As a result, many companies are falling behind in building ethical, privacy-first tools. Nearly 80% of data scientists in the U.S. are male and 66% are white, which shows an inherent lack of diversity and demographic representation in the development of automated decision-making tools, often leading to skewed data results.

Significant improvements in design review processes are needed to ensure technology companies take all people into account when creating and modifying their products. Otherwise, organizations can risk losing customers to competition, tarnishing their reputation and risking serious lawsuits. According to IBM, about 85% of IT professionals believe consumers select companies that are transparent about how their AI algorithms are created, managed and used. We can expect this number to increase as more users continue taking a stand against harmful and biased technology.

So, what do companies need to keep in mind when analyzing their prototypes? Here are four questions development teams should ask themselves:

Have we ruled out all types of bias in our prototype?

Technology has the ability to revolutionize society as we know it, but it will ultimately fail if it doesn’t benefit everyone in the same way.

To build effective, bias-free technology, AI teams should develop a list of questions to ask during the review process that can help them identify potential issues in their models.

There are many methodologies AI teams can use to assess their models, but before they do that, it’s critical to evaluate the end goal and whether there are any groups who may be disproportionately affected by the outcomes of the use of AI.

For example, AI teams should take into consideration that the use of facial recognition technologies may inadvertently discriminate against people of color — something that occurs far too often in AI algorithms. Research conducted by the American Civil Liberties Union in 2018 showed that Amazon’s face recognition inaccurately matched 28 members of the U.S. Congress with mugshots. A staggering 40% of incorrect matches were people of color, despite them making up only 20% of Congress.

By asking challenging questions, AI teams can find new ways to improve their models and strive to prevent these scenarios from occurring. For instance, a close examination can help them determine whether they need to look at more data or if they will need a third party, such as a privacy expert, to review their product.

Plot4AI is a great resource for those looking to start.

4 questions to ask when evaluating AI prototypes for bias by Ram Iyer originally published on TechCrunch

Grazzy wants to stop letting people use ‘no cash’ as an excuse to avoid tipping

Carrying cash used to be a thing, but now with credit cards, and more recently digital wallets, having more than a couple of dollars in your physical wallet is hard to come by.

Unfortunately, that also leaves many of us ill prepared to show gratitude, especially when traveling, to give a cash tip to the people cleaning our rooms or bringing your car around.

Austin-based Grazzy wants to change that through its instant pay and tax compliance platform where people can leave tips for frontline workers that employees can access on the same day. It also provides a recruitment and retention tool for service-based employers, like hotels, bars, restaurants and salons, while also giving employees financial wellness tools.

Russell Lemmer, Grazzy’s founder and CEO, told TechCrunch he was one of those travelers who experiences great services at hotels but uses his phone for everything and rarely carries cash.

“While staying in Las Vegas, I left my bags with the valet, and wanted to tip, but couldn’t,” Lemmer recalls. “I heard the valet tell his colleagues that ‘this is seven in a row.’ I felt ashamed and started to think about a solution.”

Russell Lemmer, founder and CEO of Grazzy Image Credits: Grazzy

In surveying others, Lemmer found that he was not alone: A majority of people he talked to were also heavy phone wallet users who were once in a position of not being able to tip because they didn’t have cash.

He started working on Grazzy in September 2021, and what resulted is an app that takes a “business-to-business approach to Venmo,” and a seamless way to show gratitude to someone they don’t know personally, he said.

Lemmer also wanted to help employers reduce hourly worker turnover, which he said is a $100 billion problem in the U.S., by providing a way for employees to earn and save more. He’s not alone: As the global pandemic exasperated the already tough conditions for frontline workers, other startups brought in technology to solve certain aspects. For example, Anthill to connect those who never sit at desks, SnapShift to handle HR and AskNicely for customer experience.

Here’s how Grazzy works: Guests can instantly tip staff using a property-branded QR code, and the Grazzy Direct feature allows staff to access their tips instantly. Grazzy makes money from the processing fees.

The platform also monitors and tracks the money so that businesses can be tax compliant while also seeing the wage increases and the effect on their employees.

Lemmer says Grazzy reduces the wait that often happens with tips coming through credit card transactions and the need to regularly tip out or cut weekly paychecks. In addition, by offering an alternative to traditional cash, Lemmer sees digital tips increasing an employee’s earnings by 20% on average.

Grazzy started working with its first large hotel customer last July and that has now grown to about a dozen. It is still in its early stage and started bringing in revenue in the last month, Lemmer said.

Today the company announced $4.25 million in seed funding led by Next Coast Ventures and Tuesday Capital. This brings its total funding to date to $6.8 million.

Lemmer intends to deploy the new capital into accelerating customer growth across additional hotel brands, operating groups, restaurant groups and salons and into technology development so that Grazzy integrates with major operating systems.

In the meantime, he has a three-year plan in the works that involves going after hotel operating groups that manage hundreds of properties and building out additional financial wellness features for workers.

“We want to start to layer in better ways to save and spend,” Lemmer said. “We feel that is the next step after helping them make more money and access more of it on the same day. Long-term, they stay in these jobs a little bit longer, and it is a recruiting tool for the employers.”

Grazzy wants to stop letting people use ‘no cash’ as an excuse to avoid tipping by Christine Hall originally published on TechCrunch

Shadow acquires Android emulation startup Genymobile

Shadow is making its first acquisition as it announced that it would snatch Genymobile, the company behind Genymotion. Shadow is better known for its cloud computing service that works particularly well for cloud gaming. It also offers a cloud storage service based on Nextcloud.

As for Genymobile, the French startup has been around for more than a decade. It has specialized in low-level Android development. And in particular, it has developed a popular Android emulator so that developers can test their apps on multiple configurations and following different scenarios.

Terms of the deal are undisclosed. Genymobile’s co-founder and CTO Arnaud Dupuis will stay at the company and act as the chief executive of Genymobile starting March 1st. Genymobile’s existing CEO Tim Danford will step back from the company’s day-to-day activities and move to an advisor role.

“We are very happy to announce this acquisition as the relationship built this past year with Genymobile has been pivotal to our progress in the development of the next-coming generation of our service,” Shadow CEO Eric Sèle said in a statement. “This acquisition will bring additional expertise to Shadow, and also showcases our ambition as a company, which will go through both internal and external growth.”

Genymotion started as a desktop emulator for Android development. Companies could pay a subscription price to run virtual devices on their computer. More recently, the company started offering a hosted version of Genymotion emulation. Genymobile takes care of the server infrastructure while you can use your web browser to interact with your app.

Enterprise clients can also use Genymotion to run Android virtual devices on their preferred cloud platform — Amazon Web Services, Google Cloud Platform, Microsoft Azure, Alibaba Cloud, Oracle Cloud or on premise.

Moving to the cloud means that you can use Genymotion for automated testing and trigger those tests with your continuous integration system. For instance, every time you create a new build, CircleCI or GitHub Actions can kick off some tests on Genymotion to make sure that your new version doesn’t break anything.

Developers receive an alert if there’s something wrong. In addition to Android development, quality assurance teams can use Genymotion to reproduce bugs on specific devices in a specific geographic location.

Shadow recently launched its business service for cloud computing. Companies can access high-performance virtual machines that run Windows Server. The service could be helpful for gaming studios, 3D animation companies and other industries that require powerful GPUs.

With the acquisition of Genymobile, Shadow will be able to offer access to another type of machine in the cloud — in that case, Android devices.

Shadow acquires Android emulation startup Genymobile by Romain Dillet originally published on TechCrunch

EU watchdogs agree on how to handle certain cookie consent dark patterns

Cookie consent banners that use blatant design tricks to try to manipulate web users into agreeing to hand over their data for behavioral advertising, instead of giving people a free and fair choice to refuse this kind of creepy tracking, are facing a coordinated pushback from the European Union’s data protection regulators.

A taskforce of several DPAs, led by France’s CNIL along with Austria’s authority, has spent many months on a piece of joint-work analyzing cookie banners. And in a report published this week they’ve arrived at some consensus on how to approach complaints about certain types of cookie consent dark patterns in their respective jurisdictions — a development that looks set to make it harder for deceptive designs to fly around the EU.

The taskforce was convened in response to hundreds of strategic complaints, filed between 2021 and 2022 by the European privacy rights group, noyb — which developed its own tool to help automate analysis of websites’ cookie banners and generate reports and complaints (a smart trick by a small not-for-profit to scale its strategic impact).

Cookies and other tracking technologies fall under the EU’s ePrivacy Directive, which means oversight of cookie banners is typically decentralized to regulators in Member States. That in turn means there can be varying applications of the rules around the bloc, depending on where the website in question is hosted. (Regulators in some Member States, for example, allow news sites to offer users a choice between accepting ad tracking to gain (free) access to the content or paying for a subscription to get access without tracking — although such ‘cookie consent paywalls’ remain controversial and are unlikely to pass muster with every DPA.)

Given the degree of consensus reported by the taskforce, it suggests there will be some harmonization in how DPAs enforce complaints related to the design of cookie consent banners — with, for example, the vast majority of authorities agreeing that the lack of a ‘refuse all’ option at the same level as an ‘accept all’ button is a breach of ePrivacy. So more enforcement against sites that try to bury an option to refuse tracking looks likely.

The taskforce also agreed that consent flows which feature pre-checked options (i.e. as another tactic to try to nudge agreement) is not valid consent either — which should surprise no one given Europe’s top court already clarified a need for active consent for tracking cookies all the way back in 2019.

Over the past five or so years, since another EU law came into application bolstering the rules around consent — namely the General Data Protection Regulation (GDPR) — DPAs have certainly been paying more attention to cookie consents. Including as complaints over how routinely the rules were being flouted piled up.

This in turn has led many to update (and tighten) their guidance on this issue — making it harder for sites to claim the rules around tracking consent are unclear.

Enforcements have also been picking up, with certain watchdogs being very active — such as France’s CNIL which, since 2020, has fined a raft of tech giants (including Amazon, Google, Meta, Microsoft and TikTok) for a variety of cookie-related breaches, including multiple enforcements (and fines) over the use of dark patterns to try to manipulate consent.

The CNIL’s enforcement activity has also featured corrective orders that have helped forced some major design changes — including Google revising the cookie banner it displays across the whole EU last year to (finally) feature a top-level ‘refuse all’ option. Which is quite the win.

And given the CNIL has had a leading role in coordinating the taskforce’s work, it appears that some of its convention is rubbing off on fellow DPAs.

In a press release to accompany the European Data Protection Board’s adoption of the taskforce’s report earlier this week and summarize the outcome, the French regulator writes: “This report notably states that the vast majority of authorities consider that the absence of any option for refusing/rejecting/not consenting cookies at the same level as the one provided for accepting their storage constitutes a breach of the legislation (Article 5(3) of the ePrivacy Directive). The CNIL had already taken such a position in its guidelines and in the context of several sanctions,”

As well as agreement on the need for an ‘accept all’ button to be accompanied by a ‘refuse all’ one, the taskforce agreed that the design of cookie banners needs to provide web users with enough information to enable them to understand what they are consenting to and how to express their choice.

And that cookie banners must not be designed in such a way as to give users “the impression that they have to give a consent to access the website content, nor that clearly pushes the user to give consent”, as the report puts it.

They also agreed on some examples of cookie designs that would not lead to valid consent — such as where the design is such “the only alternative action offered (other than granting consent) consists of a link behind wording such as ‘refuse’ or ‘continue without accepting’ embedded in a paragraph of text in the cookie banner, in the absence of sufficient visual support to draw an average user’s attention to this alternative action”; or where “the only alternative action offered (other than granting consent) consists of a link behind wording such as ‘refuse’ or ‘continue without accepting’ placed outside the cookie banner where the buttons to accept cookies are presented, in the absence of sufficient visual support to draw the users’ attention to this alternative action outside the frame”.

So basically they got some consensus on ruling out certain common cookie banner dark patterns.

But on visual tricks — such as the use of highlight colors which might be selected to draw the eye to an ‘accept all’ button and make it harder to see a refuse option, the taskforce decided that case-by-case analysis of the look and feel (and the potential for these kind of design choices to be obviously misleading) would be needed in most cases. And they agreed it’s not their place to impose a general banner standard (vis-a-vis colour and/or contrast) on data controllers.

They also agreed that refuse all buttons that are designed in such as way as to render the text “unreadable to virtually any user” could be “manifestly misleading” for users.

Other issues the taskforce grappled with included a more recent addition to cookie consent hell — in which sites may seek to (also) to claim a “legitimate interest” for ads processing. Sometimes adding a bunch of additional toggles alongside the consent legal basis buttons that are typically displayed only in a secondary (or other sub-menu), and where the top level does not offer a ‘refuse all’ option — instead requiring users to click through into settings to unearth this confusing mess of toggles (sometimes with the LI ones pre-checked).

“The integration of this notion of legitimate interest for the subsequent processing ‘in the deeper layers of the banner’ could be considered as confusing for users who might think they have to refuse twice in order not to have their personal data processed,” the report observes on this.

The taskforce also agreed on how regulators should determine whether any subsequent processing based on cookies is lawful — saying this would entail determining whether “the storage/gaining of access to information through cookies or similar technologies is done in compliance with Article 5(3) ePrivacy directive (and the national implementing rules) — any subsequent processing is done in compliance with the GDPR. 24”.

“In this regard, the taskforce members took the view that non-compliance found concerning Art. 5 (3) in the ePrivacy directive (in particular when no valid consent is obtained where required), means that the subsequent processing cannot be compliant with the GDPR 5. Also, the TF members confirmed that the legal basis for the placement/reading of cookies pursuant to Article 5 (3) cannot be the legitimate interests of the controller,” they add in the report.

Although they appear to have largely reserved judgement on how to handle the fresh scourge of LI toggles appearing in cookie consent flows — saying they “agreed to resume discussions on this type of practice should they encounter concrete cases where further discussion would be necessary to ensure a consistent approach”.

The working group also discussed what to do about sites that seek to classify some non-essential data processing as strictly necessary/essential — and thereby bundle it into a category which does not require consent under ePrivacy or the GDPR. However they took the view that there are practical difficulties in determining which processing is strictly necessary.

“Taskforce members agreed that the assessment of cookies to determine which ones are essential raises practical difficulties, in particular due to the fact that the features of cookies change regularly, which prevents the establishment of a stable and reliable list of such essential cookies,” they wrote. “The existence of tools to establish the list of cookies used by a website has been discussed, as well as the responsibility of website owners to maintain such lists, and to provide them to the competent authorities where requested and to demonstrate the essentiality of the cookies listed.”

On another issue — of withdrawing consent — they agreed website owners should put in place “easily accessible solutions allowing users to withdraw their consent at any time”, giving the example of a small icon (“hovering and permanently visible”) or a link “placed on a visible and standardized place”.

However they again shied away from imposing a specific standardized way for users to withdraw consent on site owners, saying that they could only be required to implement “easily accessible solutions” once consent has been collected.

“A case-by-case analysis of the solution displayed to withdraw consent will always be necessary. In this analysis, it must be examined whether, as a result, the legal requirement that it is as easy to withdraw as to give consent is fulfilled,” they added. 

EU watchdogs agree on how to handle certain cookie consent dark patterns by Natasha Lomas originally published on TechCrunch

Google parent Alphabet cuts 6% of its workforce impacting 12,000 people, saying it had ‘hired for a different economic reality’

Alphabet, parent holding company of Google, has announced that it’s cutting around 6% of its global workforce.

In an open letter published by Google and Alphabet CEO Sundar Pichai, the narrative followed a similar trajectory to that of other companies that have downsized in recent months, noting that the company had “hired for a different economic reality” than what it’s up against today. Put simply, it had bolstered its workforce during the pandemic-driven digital boom times, but it’s now having to reverse course as the world curtails its spending in the face of economic headwinds.

“We’ve undertaken a rigorous review across product areas and functions to ensure that our people and roles are aligned with our highest priorities as a company,” Pichai wrote, adding that the layoffs will impact units across Alphabet, not just Google, and that all regions and product areas will be affected.

The news means that four out of the five so-called “big tech” firms have now announced significant redundancies in the past few months, with Apple the only one of the big-five U.S. tech giant not to announce layoffs as of yet.

Indeed, earlier this week, Microsoft announced 10,000 job cuts, affecting nearly 5% of its workforce, which followed Amazon’s move to cut 18,000 jobs, or 1.2% of its global headcount. Facebook’s parent Meta, meanwhile, revealed 11,000 layoffs back in November, hitting 13% of its workforce.

Other tech giants to announce significant layoffs in recent times include enterprise software giant Salesforce, which confirmed that it was cutting 10% of its workforce at the turn of the year, impacting more than 7,000 employees.

It’s worth noting that Alphabet hasn’t been impervious to downsizing before now. Its robot software offshoot Intrinsic announced it was laying off 40 workers last week, equating to 20% of its headcount, while its life sciences subsidiary Verily scaled back by 15%, representing around 240 people.

But today’s announcement will see roughly 12,000 roles worldwide at the company disappear.

In terms of the severance, what Alphabet is offering U.S. employees seems fairly generous at first glance. The company said that packages will “start at” 16 weeks salary, plus an additional two weeks for every year worked, while they will be paid in full for the entire notification period which is a minimum of 60 days. Moreover, it committed to paying all outstanding 2022 bonuses and unused vacation time, with 6 months’ healthcare and additional support services available.

Outside the U.S., Pichai simply said that the company would “support employees in line with local practices.”

Google parent Alphabet cuts 6% of its workforce impacting 12,000 people, saying it had ‘hired for a different economic reality’ by Paul Sawers originally published on TechCrunch

Twitter puts more emphasis on the bookmark feature on iOS

After promising a feature that lets you easily bookmark tweets, Twitter is now rolling it out on iOS. The new design displays the bookmark button under the expanded tweet view. Making it easier to add a post to your bookmarks.

Before today’s change, you had to tap on the share button to open the sharing card and then tap on the bookmark option to save a tweet. In addition to the new button, as soon as you tap on the button, you will see a banner at the top of the screen that says “Show all bookmarks.”

The option is currently visible only on the iOS app, but we can expect that Twitter will roll this out to Android and the web soon.

Image Credits: Twitter

Earlier this month, Elon Musk promised to bring the bookmark button to the detailed tweet view. Now, it is finally rolling out to end users. Many people have used the “Like” button as a bookmark feature — just because it was cumbersome to save and retrieve tweets as bookmarks. But this new button placement might change things a bit.

Easy swipe right/left to move between recommended vs followed tweets rolls out later this week.

First part of a much larger UI overhaul.

Bookmark button (de facto silent like) on Tweet details rolls out a week later.

Long form tweets early Feb.

— Elon Musk (@elonmusk) January 8, 2023

Notably, if you have a Twitter Blue subscription — which is now also available on Android — you can use the bookmark folders feature to arrange them better.

Along with this, Twitter has also updated its Android app to show the two timeline options in two side-by-side separate tabs. The company first rolled out the default “For You” timeline and chronological “Following” timeline to iOS last week.

It’s your turn Android — no more icon. Upgrade to the latest version of the app to switch between “For you” and “Following”.

— Twitter Support (@TwitterSupport) January 20, 2023

Later it brought this view to the web as well. At that time, Twitter remembered your choice of timeline even if you closed the tab. So for instance, if you chose “Following” and opened Twitter in a new tab after closing it, the “Following” timeline appeared as the default feed. However, the company now seems to have made the change in such a way that you will see the “For You” feed by default if you refresh the page or open Twitter in a new tab.

Meanwhile last night, the company made changes to its API terms, effectively killing third-party clients. As a result, well-known apps like Twitterrific have started pulling their apps from the app stores.

Twitter puts more emphasis on the bookmark feature on iOS by Ivan Mehta originally published on TechCrunch

Fandom lays off employees across Giant Bomb, GameSpot and Metacritic

Entertainment company Fandom has laid off an unspecified number of employees across multiple properties including Giant Bomb, GameSpot and Metacritic. These properties are mainly focused on publishing content around gaming and TV shows.

According to a report by Variety, the company employs around 500 people, and the layoffs have affected roughly 10% of its staff across different sites. Employees were caught off-guard by this surprise announcement made by CEO Perkins Miller during an all-hands meeting, as per Kotaku.

The job cuts come months after Fandom acquired a host of brands — Comic Vine, Cord Cutters News, GameFAQs, GameSpot, Giant Bomb, Metacritic, and TV Guide — from Red Ventures in October. Sources told TechCrunch at that time that the size of the deal was around $55 million.

During the acquisition announcement, Fandom said that this deal will help the company “super-serve” its advertising partners and provide fuel for its data platform and gaming e-commerce verticals. Fandom was founded in 2004 by Wikipedia co-founder Jimmy Wales and entrepreneur Angela Beesley

Some folks like GameSpot entertainment editor Mat Elfring, video producer Jess O’Brien aka “Voidburger”, and graphic designer Justin Vachon tweeted about frustrating layoffs.

Whiskey Media (Comic Vine)->CBS Interactive (Comic Vine/GameSpot)->Red Ventures (GameSpot)->Fandom (GameSpot)->Jobless

I love you all at @GameSpot. You all made me so happy. That job was something I was so luck to have for 7 years. I cannot believe I’m gone.

— Mat Elfring (@ImMatElfring) January 19, 2023

When your company lays off 2 of the 3 editors on staff lmao pic.twitter.com/QCJPEGvO8D

— VoidBurger (AKA Jess) (@VoidBurger) January 19, 2023

Well, for the first time in my life ive been laid off and…just feel completely numb. Its been a wonderful year+ working with @GameSpot @giantbomb @TVGuide and @metacritic but now i gotta start to figure out what the fuck is next, just at a loss for now to say the very least

— justin vachon (@megaberrycrunch) January 19, 2023

We have reached out to Fandom for a comment, and we will update the story if we hear back.

Several gaming and entertainment media outlets have faced layoffs in the last few months. Some IGN staffers were impacted last month while Polygon employees were axed as a part of Vox job cuts in July.

Fandom lays off employees across Giant Bomb, GameSpot and Metacritic by Ivan Mehta originally published on TechCrunch

India releases guidelines for social media influencers accepting paid promotions

As the market of social media influencers is getting bigger in India, the South Asian nation has introduced endorsement guidelines to limit unfair trade practices and misleading promotions on the web.

On Friday, the Department of Consumer Affairs held a press conference to announce new guidelines to make it mandatory for social media influencers to disclose promotional content in accordance with the Consumer Protection Act, 2019.

Failing to follow the guidelines will make social media influencers liable for a fine of up to $12,300 (1 million Indian rupees). In the case of repeated offenders, the penalty can go up to $61,600 (5 million Indian rupees), the Indian government department said.

The department specified that the guidelines apply to social media influencers as well as virtual avatars promoting products and services online. The disclosures should be easy to notice in post descriptions where you can usually find hashtags or links. It should also be prominent enough to be noticeable in the content, the department said.

When it comes to promoted content in videos, the department said that disclosures for paid promotions should be placed in the video — not just in the description — and be made in both audio and video format. Influencers must also disclose if they promote a brand, service or product during live streams, per the guidelines.

The department said the disclosures and endorsements should be in the language of the content.

“Today’s guidelines are aimed at social influencers who have a material connection with the brand they want to promote on various social media platforms. So this is an obligation for them to behave responsibly,” consumer affairs department secretary Rohit Kumar Singh told reporters.

The official said the department was in talks with tech companies to deploy some crawling algorithms to identify offenders. Meanwhile, consumers can file complaints if they find an influencer violating the guidelines, the secretary said.

“You can never cover it 100%. This is a cat-and-mouse thing… So, the idea is to protect the interests of the consumers and let him not being taken for a ride by showing him something as unbiased whereas actually, it is a paid thing,” he added.

According to the secretary, the size of the social media influencer market in India in 2022 was $157 million. It could reach as much as $345 million by 2025.

India releases guidelines for social media influencers accepting paid promotions by Jagmeet Singh originally published on TechCrunch

Pin It on Pinterest