May 11, 2022

PROTOCOL Coverage | Newsletter | Intel | Events Are you keeping up with the latest cloud developments? Get the Enterprise team's newsletter every Monday and Thursday. Email Address Enterprise Thank you for signing up. Please check your inbox to verify your email. Email me an authentication link A login link has been emailed to you - please check your inbox. Kate Kaye is an award-winning multimedia reporter digging deep and telling print, digital and audio stories. She covers AI and data for Protocol. Her reporting on AI and tech ethics issues has been published in OneZero, Fast Company, MIT Technology Review, CityLab, Ad Age and Digiday and heard on NPR. Kate is the creator of RedTailMedia.org and is the author of "Campaign '08: A Turning Point for Digital Media," a book about how the 2008 presidential campaigns used digital media and data. Image: Datagen Kate Kaye is an award-winning multimedia reporter digging deep and telling print, digital and audio stories. She covers AI and data for Protocol. Her reporting on AI and tech ethics issues has been published in OneZero, Fast Company, MIT Technology Review, CityLab, Ad Age and Digiday and heard on NPR. Kate is the creator of RedTailMedia.org and is the author of "Campaign '08: A Turning Point for Digital Media," a book about how the 2008 presidential campaigns used digital media and data. May 11, 2022 Sloshed in the subway after a night of partying? Bored during a virtual meeting? Dozing off at a red light? Companies are building software that uses AI to monitor people’s behavior and interpret their emotions and body language in real life, virtually and even in the metaverse. But to develop that AI, they need fake data, and startups are stepping in to supply it. Synthetic data companies are providing millions of images, videos and sometimes audio data samples that have been generated for the sole purpose of training or improving AI models that could become part of our everyday lives in controversial forms of AI such as facial recognition, emotion AI and other algorithmic systems used to keep track of people’s behavior . While in the past companies building computer vision-based AI often relied on publicly available datasets, now AI developers are looking to customized synthetic data to “address more and more domain-specific problems that have zero data you can actually access,” said Ofir Zuk, co-founder and CEO of synthetic data company Datagen. Synthetic data companies including Datagen, Mindtech and Synthesis AI represent a corner of an increasingly compartmentalized AI industry. They produce AI parts that will eventually be assembled to build software, features in applications or systems used in vehicles. They serve customers such as computer vision engineers and data scientists working for big tech giants, automakers, gaming companies or mobile phone makers. Image: Datagen Like so much polyester, synthesized datasets are intended to mimic the real thing. Synthetic data does not just replicate actual photo and video data; it enhances it by adding dimensions and details that help AI-based systems learn. Sometimes the synthetic stuff fills serious data gaps where real data does not exist or is difficult to obtain. It might depict dangerous highway situations used to train autonomous vehicle AI, or include facial images representing people of multiple ethnicities or ages needed to help ensure AI makes fair and accurate decisions. Many of these companies tout synthetic data as a panacea for the lack of diverse AI training datasets that has contributed to discriminatory AI, particularly facial recognition . “We help customers reduce AI bias by providing synthetic data spanning a wide range of age, gender, BMI and ethnicity,” said Yashar Behzadi, CEO of Synthesis AI. Training AI to spot drunks and cheats “One company came to us recently needing a solution to detect cheating in exams,” said Steve Harris, CEO of Mindtech, a company that offers a platform for designing and rendering images based on photorealistic computer graphics. Harris would not name the virtual testing technology customer, but said that like many other virtual testing tech companies , the customer wanted to incorporate AI to monitor whether test takers are showing signs of cheating such as looking away from the computer screen or interacting with a person or a phone. For an AI model to pick up on all the different possible signs of cheating in multiple environments involving a variety of people, it would need a large corpus of imagery showing hand, eye and body movements to learn from — the sort of images that could be too expensive to purchase, or force violations of privacy to obtain, even if there were enough of them. “It becomes even more complex when you throw in facial key point data and skeleton pose data to train systems to understand which way the student’s gaze is going, which way their body is about to turn or which direction their hands are facing,” Harris said. The Center for Democracy and Technology reported that AI-based systems used to detect cheating on virtual testing pose risks to student privacy and mental health , and can discriminate against disabled people. Harris said customers are also interested in using synthetic data to build AI for use in public places like transportation hubs. Some have sought data to train AI systems to prevent an impaired person from getting hurt in a train station, for instance, “if someone is drunk and moving too close to the line,” he said. Mindtech recently unveiled a package of pre-built images designed specifically for retail environments that could be used to track customers’ interest in specific products, keep track of stock on store shelves or predict traffic flow in parking lots. In the earlier days of the pandemic, Mindtech customers wanted synthetic data to train AI to monitor for compliance with face mask rules, Harris said. As companies continue to develop AI for a never-ending array of applications, investors see a bright future for the synthetic data makers supplying the raw materials. Mindtech collected $3.25 million in funding last year. Synthesis AI received $17 million in series A financing in April, and Datagen gathered $50 million in a series B funding round in late March. Facial expression data detects meeting boredom and driver distraction For now, a lot of the AI that synthetic data helps build is used in mundane, real-world situations such as driving. Synthetic data is training AI models for driver monitoring systems that capture driver images through dashboard cameras and use computer vision AI to detect distraction, such as in delivery vehicles . “We have a number of customers in that area,” Behzadi said. For example, Synthesis AI provides synthetic facial data to Affectiva, a company that offers AI-based systems to estimate people’s emotions and cognitive states in real time to detect driver distraction and behavior such as road rage, according to Rana el Kaliouby, founder of Affectiva. “The data we provide Affectiva and others in driver monitoring is focused on improving driver safety — actions like falling asleep, not wearing a seatbelt or being distracted. We believe synthetic data can be a net positive in reducing fatalities and improving overall safety,” Behzadi said. AI-based systems for assessing driver behavior are bound to be more commonplace in coming years. The European Commission has made distraction and drowsiness recognition features mandatory for new vehicles this year. Meanwhile in the U.S., the 2022 Infrastructure Act set aside funds to study the use of driver-monitoring systems to minimize driver distraction. Synthetic data is also fueling AI for workplace monitoring. Datagen’s latest product lets customers build annotated, 30-frame-per-second images for use in office, meeting and conference technologies. Zuk said the new data might be used to train AI that detects whether someone is bored during a work meeting; it includes image data showing people holding their heads in their hands, for instance. Image: Datagen Already, AI-based features for assessing people’s emotional states are showing up in virtual classroom platforms and even sales meeting software , although synthetic data has not been used to train many of these systems yet. Human rights advocates are fighting potential use of emotion AI in everyday tech. More than 25 organizations including the American Civil Liberties Union, Electronic Privacy Information Center and Fight for the Future sent a letter on Wednesday to Eric Yuan, founder and CEO of Zoom, demanding the company end plans to incorporate emotion AI in its software features. The letter and previous efforts by Fight for the Future were prompted by reporting in Protocol in April about Zoom’s potential plan to incorporate emotion AI into its products. “This software is discriminatory, manipulative, potentially dangerous and based on assumptions that all people use the same facial expressions, voice patterns, and body language,” the groups wrote. Mindtech’s Harris said companies building the next wave of office and meeting software are interested in monitoring human interactions to “flag things that look unusual.” He said he expects companies like Meta and Google to incorporate this type of tracking AI in the virtual environments they create, but he added, “It’s some way off.” For AI to pick up on whether people are paying attention to the road — or to the boss during a meeting — it often needs to recognize facial expressions. Synthesis AI’s datasets include minute distinctions among millions of images expressing as many as 150 facial “micromovements,” Behzadi said. Customers use the company’s digital system to submit requests for custom data, then it automatically renders what they ordered. “They’ll say, ‘I need a million images that span all these different dimensions,’” Behzadi said. The result might be thousands of facial images with a variety of skin tones, hair styles or features like hats or glasses. Diversity also manifests in the way images are lighted, Behzadi said. If Synthesis AI is making data for use in an augmented reality environment, the system will produce multiple versions of images of people or objects — a mug of coffee on a desk, for example — with variations in the direction that light emanates from. “So when I render this image in the scene it’s realistic,” he said. A step removed from ethical implications According to el Kaliouby, Affectiva has used synthetic data to increase the diversity of its dataset representing people across age ranges and ethnicities. Merely having lots of faces across cultures may not be enough to train AI that also needs to learn what people look like when they are wearing a ball cap, when they are alert or asleep or in environments with low or bright light. “It gets really complex and expensive super fast to scale this,” she said. But as synthetic data companies push a diversity mission, their products may be used to build contentious forms of AI. The legitimacy of emotion AI has been questioned by researchers who say neither humans nor machines can accurately detect people’s emotions based on facial expressions. And in general, many also believe that algorithmic systems monitoring people’s facial expressions or how they walk or talk perpetuate unnecessary surveillance and could be used to unfairly penalize people. However, in some cases, synthetic data suppliers remain a step removed from the products that will be manufactured using their data. Image: Datagen Because it is automatically produced, synthetic data comes with some baked-in metadata including details about what images and videos represent that are necessary to help AI models learn. In Datagen’s case, the company includes labels showing the intensity of facial expressions such as “slightly smiling” or “extremely happy,” but also lets customers add custom labels. “Datagen is propelling advancements in AI by removing the need to source and manually annotate training data,” said Gil Elbaz, Datagen’s CTO and co-founder. The company did not say whether it has restrictions on the types of customers it will supply data to, other than to say it is “focused on commercial Computer Vision applications enabling AI teams to develop safe, human-centric use cases.” But other synthetic data companies leave some of the data-labeling decisions to their customers and stay at arm’s length from the end products they help make. Instead of providing qualitative labels categorizing facial expressions as confused or bored, Synthesis AI only annotates facial images with technical information. An image label might include metadata stating that the left side of the mouth moved upwards 10 degrees, but would not come pre-labeled as “slightly happy,” for instance. While Behzadi said Synthesis AI has turned down work with customers that wanted to use its data to identify people without their consent, he said the company has not turned down potential customers that want data to train emotion AI models. Mindtech also leaves the labeling and end-product decisions to customers. “We understand that facial expressions are very subjective, so we allow a customer to determine how they want to use any labels,” said Chris Longstaff, Mindtech’s VP of Product Management. He said customer confidentiality prevents Mindtech from knowing details of products built with its data. Expect more synthetic data creation in the near future as it forms the foundation of all sorts of AI built for emerging virtual environments. “There is the potential for synthetic data to be a prominent tool for metaverse companies,” said Harris. Behzadi said he expects interest in metaverse-related uses for synthetic data to ramp up in the next year. Synthesis AI is working with customers that want to digitize sporting events in real time for metaverse environments. In the future, Behzadi said, “I can watch the game from Tom Brady’s eyes.” Keep ReadingShow less May 9, 2022 April 26, 2022 Imagine: You’re the leader of a real estate team at a restaurant brand looking to open a new location in Manhattan. You have two options you’re evaluating: one site in SoHo, and another site in the Flatiron neighborhood. Which do you choose? Companies that need to make these types of decisions leverage foot traffic patterns coupled with additional data sources to build a sound approach to real estate and investment decisions. Below, we take a closer look at points of interest and foot traffic patterns to demonstrate how location data can be leveraged to inform better site selecti­on strategies. Analyze: Make sense of where people are moving to inform better business decisions. Model & Forecast: Identify and predict trends based on foot traffic in different regions, cities and neighborhoods. Select sites: Determine where to place new locations or develop properties based on foot traffic (or lack thereof) in commercial districts. Derive insights: Deeply understand points of interest and behavioral patterns, and how they're changing over time. Here’s how foot traffic data can impact site selection or real-estate decisions. Look at your competitive set: Identify current venues in a neighborhood or area to determine where there might be white space and to quantify the competitive landscape. Analyze your overall competitive set (e.g., in this report we looked at all restaurants) as well as more specific, relevant categories of venues (e.g., in this report we looked at cafes). Know which places your prospective customers go now, and where you might have an opportunity to take market share or position yourself alongside businesses that provide synergies. Know whether your consumer traffic would come from tourists, or locals: Classify tourists versus locals by looking at individuals with home ZIP codes more than 120 miles away in your analysis to better understand the catchment area (i.e., where consumers are coming from). Know more about consumers in your neighborhood: Analyze the demographics of consumers in a particular neighborhood to understand the types of people a prospective site might draw so that you can select the optimal location based on your target audience. Uncover changes in visit patterns over time, and within a typical week: Look at a particular neighborhood over time in order to capitalize on trends, selecting a site where traffic may be on the rise. Compare visitation patterns by neighborhood to understand the traffic you might expect to see throughout the week at a given site, informing and validating (or invalidating) your projections. Know what day of week experiences the most natural footfall traffic. Understand the trends and what your consumers like: It’s critical to know what consumers are looking for, how they spend their time and what they like now and into the future. Use data-visualization platforms and tools to make insights easy: Data-visualization platforms make complex information and insights easier to understand and ultimately react to. You’ll see companies that adopt data visualization are empowered and can spot emerging trends and speed reaction time. We’ve demonstrated the benefits of using foot traffic data to in a use case that evaluates data to determine whether to build a restaurant in SoHo or the Flatiron neighborhood. For this analysis, we aggregated Census Block Groups into Census Tracts to define and analyze Manhattan’s SoHo and Flatiron neighborhoods. By leveraging the new home and work CBG attributes, we were able to provide a more granular understanding of where consumers live and work to inform business analysis and decisions. This new level of detail allows you to layer census data such as demographics onto your analysis in order to learn more about visitors to categories, chains and venues of interest. Foursquare analyzes consumer behavior based on foot traffic data from millions of Americans that make up our always-on panel. For the purpose of this report, all data is anonymized, aggregated and normalized against U.S. census data to remove any age, gender and geographical bias. Key learnings: Different target audiences with different needs SoHo: Consumers visiting restaurants in SoHo are primarily locals (83%) ages 25-34 (44%). Restaurants in this area attract super shoppers, affluent socialites, health-conscious consumers and a cultured and artsy crowd. Flatiron: People visiting restaurants in Flatiron are primarily locals (86%) ages 25-34 (46%). Restaurants in this area attract health-conscious consumers, corporate professionals, college students and people who crave unique experiences. Visitation patterns and staffing/hours of operation vary Soho: A restaurant in SoHo may struggle to draw consistent foot traffic throughout the earlier part of the day and week: Restaurants in SoHo rely heavily on weekend visits (38% of total weekly visits) in the late afternoons (60% of total daily visits occur after 3 p.m.). Flatiron: A restaurant in Flatiron may struggle to draw consistent foot traffic throughout later day-parts and weekends: Restaurants in Flatiron rely heavily on weekday visits (70% of total weekly visits) in the earlier part of the day (45% of total daily visits occur before 3 p.m.). Competitive differences SoHo: A new restaurant in NYC's SoHo neighborhood will face tough competition with more than 435 restaurants in the area, including over 48 cafes. Top-visited restaurants in this area include Gitano, Prince Street Pizza and Thai Diner. Flatiron: A new restaurant might face less competition in Manhattan's Flatiron neighborhood, with roughly 267 restaurants in the area, including only 25 cafes. Top-visited restaurants in this area include Eataly, Shake Shack and The Smith. While a new restaurant in NYC's Flatiron neighborhood may face less competition compared to a new restaurant in SoHo, location data verifies what it takes to be successful in both neighborhoods. Outcomes and next steps In order to be successful in Flatiron, a restaurant will need to draw a weekday lunch crowd with healthy offerings and a work-friendly setting for professionals; to stand out among nearly double the restaurants in SoHo, a new restaurant should lean into arts and culture with a design-forward setting, and focus on evening and weekend offerings. Read the full report to better understand the role of location data in uncovering trends in consumer behavior, assessing the competitive landscape and unlocking unique opportunities for venue expansion. Keep ReadingShow less Amber Burton (@ amberbburton ) is a reporter at Protocol. Previously, she covered personal finance and diversity in business at The Wall Street Journal. She earned an M.S. in Strategic Communications from Columbia University and B.A. in English and Journalism from Wake Forest University. She lives in North Carolina. May 11, 2022 Rich Cho is talent acquisition platform Gem’s new CRO. But he’s not in charge of revenue. He’s the chief recruiting officer. On the heels of an all-out brawl for high-skilled tech talent, Cho is beginning amid a cooling in the market . A growing list of tech companies are announcing they will not only slow down hiring, but freeze it as the market continues to slump and layoffs continue . But Cho says he’s not worried. The 20-plus-year recruiting veteran has worked through a total of three economic downturns over the course of his career, which has spanned some of the hottest hyper-growth tech companies including eBay, Facebook and, most recently, Robinhood. He spoke with Protocol about why now it is still more important than ever to have someone in leadership focused on recruiting and hiring. Cho advises that slowing down can produce a more thoughtful and productive hiring process, but companies should still be as responsive to candidates as possible. And when it comes to hiring at a startup, recruiting is the most strategic thing you can do right now. This interview has been edited for brevity and clarity. What attracted you to working at Gem and jumping into this role that's similar to your prior roles, but has a very different title? One thing you'll find about companies at an early stage is that recruiting is the most strategic thing they can do when they don't have a brand and they're really building out the organizations from the ground up. So recruiting leaders, like myself, are one of the most strategic executives on the team. And over time, as a brand gets established and the recruiting flywheel gets moving, recruiting then becomes more of a service organization. What drove me to Gem is that Gem had said that recruiting will remain strategic in perpetuity, as long as we're around and growing. That's something that was the most attractive to me. What are the responsibilities and new focuses of a chief recruiting officer? The new types of responsibilities tend to be around helping our executive staff make really important decisions. When it comes to things like location strategy … that is a vigorous debate that's happening in the market today. Companies like Airbnb are announcing that they're full time remote. The founder felt that that's not only the wave of the future, but that's where you're going to get the most candidates — you have more optionality with candidates. Now, we're not quite sure if that's right or wrong based on the data, but that's a really important strategic decision that the C-team has to make. And the role that's new in the CRO title is that I have an opportunity to help facilitate that discussion and come up with a proper decision that impacts the company long term. What are some of the KPIs that you're going to be paying attention to in order to know you're being effective in this role? I think it first has to start with the core metrics that allow us to determine whether or not the recruiting is moving forward in the right direction. These are simple things like conversion metrics, from when we originally engage a candidate to how successfully we're able to close them. But from that core set of data, we can now do more insightful analyses around, you know, is there a correlation between how quickly we hired them to how often they accepted an offer? There are other bits of data that are also really important in that funnel. And last but not least, a whole new field of data that allows us to then ask the really difficult questions around: Are we hiring in the same way that tech has been hiring in the past? Or do we have any insights that indicate that we can be more innovative in the selection process so that we can welcome in candidates from non-traditional backgrounds that would help to increase our company's overall diversity as a result of that? That's going to be the heart of what allows us to not only create a more efficient and effective recruiting system, but allows me in my role to facilitate proper conversations at the C-suite so that we can make decisions that impact the company's culture and the growth of the company long term. Oftentimes when it comes down to just trying to get people into a role, a lot of diversity and hiring initiatives are the first things to go out the window. From your experience, how do you keep that from happening, and what do you keep top of mind to make sure that you're still connecting with a diverse array of candidates? Speed will always have some level of detrimental impact. The main reason for that is because of the fact that the way we've been selecting [hires] has been codified for decades. We have a certain set of questions that we put a lot of trust into and so as a result that's going to produce a more homogenized outcome. What the teams that I have been a part of have attempted to do is set up an experiment to see if there are certain sets of questions that folks from traditional backgrounds — meaning the same colleges or companies that you're used to hiring from — how are they performing against those that do not come from the same colleges and companies that [you] are used to hiring from? Then you start to ask yourself what’s the efficacy of those actual questions, and if they're not effective, can you change them or replace them with something that actually is a much better indicator for success? Those things by proxy slow down the process a little bit. In today's environment, there might be an argument to be made that hiring thoughtfully is the right strategy over hiring as fast as you can in a frenzied environment. You start to bring in a selection process that, in my opinion, actually produces better results. So you start to go slow only to go faster at the back end of [the hiring process]. What do you think hiring looks like, especially as it relates to diversity, in a time when tech hiring is possibly slowing down over the next several months? Now is the absolute best time to focus on creating better processes, more objective processes, and increasing efficiency. I've been through three recessions in my professional career and as a result, I know recruiting is going to rebound. And in many cases, when it rebounds, it rebounds aggressively because the first staff to be cut tends to be on the recruiting side, and then you don't have enough resources to sort of dig out of it. But companies that actually prepare and create better efficiency during this time end up in a much more advantageous and competitive position. This is also the time when you can invest in those experiments that I just mentioned. The hardest part about actioning true change in the selection process is to apply some of the things that we're learning philosophically around unconscious bias. Applying that to the questions that we asked and then creating a structure that allows for the interview panel to be more thoughtful around how to call out when certain biases occur or certain things in the hiring process occur. And that takes time, to be able to go through that change management. But now is that time to go through those experiments, go through that full transformation, and at the same time also instrumenting more thorough [and] more efficient recruiting practices. During the 2008 recession you were at eBay and then Facebook. What lessons did you take from working in recruiting during prior downturns that you’re keeping in mind right now? It was a tale of two companies. EBay was directly impacted based on the recession that we were moving into, whereas Facebook was a less-than-500-person startup that was really gaining a lot of momentum. The key to Facebook's success was to bring in the best talent that they could to get through that very transitional period of growth. So while companies like Google were pulling back hiring and laying people off, recruiters [at Facebook] were on the back end bringing them in as aggressively as possible. So the three lessons that I'll say is that in this environment every company has an advantage to being able to bring in top talent that aligns to their mission. It requires a lot of work around building a strong talent brand message that addresses why that company would be great for any candidates aligned to their mission and values. Talk about why that company is still growing in this period of recession or slowing down hiring. And then, obviously, market that appropriately. So that's point number one. Point number two, as companies start to talk about slowing down, or freezing or laying off hiring, you're going to start to see a lot of the quality of inbound candidates increasing exponentially, because now people are saying, “Okay, well, if my company is not growing, I should probably think about a growth-based company.” So the lesson that I learned is to start to think about how to not only increase talent brand, but also work on a process that allows you to be faster and more responsive to inbound candidates. And then the third and final one, which is, this is more philosophical. I've seen during down economies people becoming more risk averse. In this period of time, we spent a lot of time at places like Facebook early on talking about why now is the time, especially in an organization that's small, to elevate your career trajectory because you're going to have an outsized impact in your role, your industry or your career … So philosophically, I would highly encourage candidates not to be as risk averse. Obviously, look through the business metrics to make sure the company that they're going to join has solid fundamentals, but now's the time to really think about taking the leap in a high-growth company. Sounds like you were not nervous at all this morning reading the headlines about the coming slowdown in the tech industry. I was really excited that we would have an opportunity for amazing talent that would be potentially looking for something new and something different than the companies that have to make some cost-cutting decisions. That being said, why is it so important with everything going on to have a chief recruiting officer in the C-suite right now? With the slew of new trends that we're seeing in talent acquisition from remote hybrid work to navigating this new economy, and understanding the role that talent brand plays in really speaking to the heart and passion of candidates, it requires someone that has a seat at the table. Someone that both will be a great counsel to the rest of the C-suite, but also help shepherd great decisions that allow companies to really take advantage of an opportunity where you now will have access to some amazing talent in the market today. And I say that for both startup companies, midsize companies or newly minted unicorns like us. But I also believe this is important for 100-year-old brands to think about how they're going to continue to attract talent in this new talent acquisition market. Keep ReadingShow less Veronica Irwin (@vronirwin) is a San Francisco-based reporter at Protocol, covering breaking news. Previously she was at the San Francisco Examiner, covering tech from a hyper-local angle. Before that, her byline was featured in SF Weekly, The Nation, Techworker, Ms. Magazine and The Frisc. May 10, 2022 We all know when AI crosses an ethical line. Automated lending systems charging higher rates for people of color? Bad. News-feed algorithms feeding diet pill ads to teenagers with eating disorders? Yeah, that’s not so good either. What’s less easy is understanding what each of these examples have in common, and drawing lessons that apply to early-stage companies. There are plenty of broad statements of AI ethics principles, but few tools for putting them into practice , especially ones tuned for the harsh realities of startups tight on money and time. That challenge extends to VCs too, who must increasingly attempt to assess whether founders have thought through how customers, partners and regulators might react to the ways they’re using artificial intelligence. Even when founders have the best intentions, it’s easy to cut corners. But without a clear ethics framework, the consequences can include regulatory delays, a longer road to profitability and even real-world harms that can do long-term damage to a company’s reputation. To solve this problem, a group of consultants, venture capitalists and executives in AI created the Ethical AI Governance Group last September. In March, it went public, and published a survey-style “continuum” for investors to use in advising the startups in their portfolio. The continuum conveys clear guidance for startups at various growth stages, recommending that startups have people in charge of AI governance and data privacy strategy, for example. EAIGG leadership argues that using the continuum will protect VC portfolios from value-destroying scandals. Anik Bose, general partner at Benhamou Global Ventures, is the executive director and founder of EAIGG. He spoke with Protocol about how startups can align their processes with their values and why he’s making sure companies in his firm’s portfolio follow the continuum’s advice. This interview was edited for brevity and clarity. How did you know that now was the right time to begin standardizing AI ethics? Hasn’t it always been important? AI is a double-edged sword. For one, it has tremendous promise across industries: manufacturing, health care, consumer products, insurance, banking, you name it. Some of that promise people are betting on: Private investments in AI are booming. If you look at the patent filings in AI, they’re skyrocketing. And if you look at the top skills sought after by employers today, No. 1 is a Ph.D. in AI. Along with that comes fear of AI. The first fear, which is very visceral, is robots replacing humans, like the Terminator. The second fear is the fear of the concentration of AI assets. If you look at the FAANG companies, there's this fear that these guys will prevent the democratization of AI, because they have all the resources, all the people and are basically doing all the acquisitions in the space. Then you look at AI policy today, and it's the Wild West. There's very little or no regulation in the U.S., it’s just now coming in Europe, and there's a lack of general awareness about things like social exclusion, privacy intrusion and discrimination. Given all that, we believe that now is the right time to operationalize AI ethics. You can't really wait for regulation to show up and tell you what to do. Why do AI ethics matter from a business perspective? It’s about customer trust and market adoption. With early-stage startups, you're going and making evangelical sales to large enterprises. If they don't trust you or your product, you're in deep doo-doo. If your AI model is doing things that it is not supposed to do, you're done. Secondly, regulation is coming. If one starts addressing this now, while they’re still a startup, they will be much more ready to handle when the guillotine drops. The other two reasons are equally important, though people often don’t get this: Attracting and retaining top talent is the No. 1 issue for startups. More and more people want to make sure that the startups they work for have a deeper purpose beyond making money. They want to make the earth a better place. You're not going to recruit talent if you're building products in a mercenary way and not dealing with these issues. Last, once you get to the place where you want to be acquired — like, let's say Microsoft or Google approaches you — I can tell you that during diligence on M&A, they're going to look at your ethics framework. If there’s some liability, not only will you not get acquired, but the valuation of your company might drop by a factor of 10. Why is it important to have a single person in charge of AI ethics, rather than just making sure all employees are on board with the company’s values? We fundamentally believe that accountability is best established by assigning clear-cut responsibility. Someone has to own it. We have learned through our experience in startups that the No. 1 reason for poor implementation of really anything is a lack of clear accountability. So we fundamentally believe that unless you assign AI governance to a person, it’s not going to get done. Think about it: The title “chief information security officer” did not exist in enterprises in the 1990s. Today, every enterprise has one. Is that one person responsible for the actions of the entire company? No, but they're ensuring the processes are in place. They're ensuring the tools are being used. At the end of the day, the board or the CEO can go to one person and say, “Where are we on this?” What should that person’s title be within the organization? In the early days it's going to be the VP of Product Management, chief product officer or the founder driving the product, because they're the ones actually building with AI. They are the guys who can figure out, “Are the right data sets being used?” or “Is there model drift?” Later, when you're creating $20 million, $50 million in revenue, you might have multiple products, and you might be using data in different ways. At that point, it makes sense to have someone in charge of just ethics, like an ethical AI officer or an adviser. You see a lot of late-stage startups today have a chief ethics officer. We believe that's going to become more common. What are the next steps toward getting tech startups on board with AI ethics? If you take a step back, education is a big part of the conversation we are having. Part of why we founded EAIGG was to open source best practices, so everyone can learn from each other. The continuum is just one tool, but we also hosted a panel discussion about what financial services are doing in terms of AI governance and what their best practices are. We had another panel discussion with IBM, where they talked about Fairness 360 , a tool they've open sourced and that we promote as a tool to use with AI models. I think the continuum is a powerful tool for startups, but what we want with EAIGG is both to do more research to create other tools and also push to open source tools that companies are already using today. I'm sure Google has got a lot of best practices that not many people know about, for example. Last, we’re also going to compile tools to help people get on board with regulation. We believe Europe will lead the way with regulation, like they did with GDPR, and that the U.S. will follow. When regulation comes on a broader scale, and people get fined $5 million, $10 million, $50 million — I can tell you that people will start paying attention to AI ethics. Keep ReadingShow less Hirsh Chitkara ( @HirshChitkara ) is a reporter at Protocol focused on the intersection of politics, technology and society. Before joining Protocol, he helped write a daily newsletter at Insider that covered all things Big Tech. He's based in New York and can be reached at hchitkara@protocol.com. May 10, 2022 Nothing disrupts policy agendas quite like a recession. That’s doubly true for a recession taking shape six months before midterms. Politicians will be eager to quell the frustration that comes with inflation, diminishing job prospects and plummeting portfolios — all of which are converging in the aftermath of a two-year lockdown that eroded public trust in the media, elected officials and the scientific establishment. The historic formula for addressing recession anger? Punch up. And by those standards, there’s no target more punchable than tech. Jeff Bezos, Mark Zuckerberg and Elon Musk all saw their wealth grow by tens of billions of dollars since March 2020. The pandemic minted a new billionaire roughly every 26 hours , with most of that wealth emerging from the tech sector. Meanwhile, most U.S. households are staring down bleak prospects: Their communities have been ravaged by opioids , home ownership is more elusive than ever and the remote-learning experiment deprived an entire generation of adequate socialization. The tech policy environment is primed to undergo a fundamental shift — a vibe shift , if you will, only swap out the cig-smoking podcasters with Tesla-driving D.C. lobbyists. Government subsidies will receive more scrutiny than ever. The crypto lobby will need to pay closer attention to the optics of policy prescriptions. And with mounting anxiety around employment prospects, tech firms will face greater regulatory scrutiny over hiring practices. One final note before diving into the specifics: The magnitude of these shifts depends on how much worse things get from here. Nasdaq still stands 80% higher than it did in March 2020. The markets rallied Tuesday morning with some investors ready to “ buy the dip .” We aren’t anywhere close to dot-com territory yet, and a few notable bears warn that this recession could make the dot-com crash look rosy. Only time will tell. Hiring under the microscope Until recently, tech companies struggled to fill open roles. That translated into a golden era for software engineers, with the labor shortage driving up compensation by around 20%. Big Tech resorted to dishing out giant bonuses to retain talent. Apple reportedly gave some engineers “special retention grants” worth over $200,000 in RSUs. Amazon raised its cash-pay cap from $160,000 to $350,000. But there are already signs of tech firms going in the opposite direction to instead trim labor costs. Last week, Meta CFO David Wehner sent an internal memo announcing hiring cutbacks that will impact “almost every team across the company" through the end of 2022. Days later, Uber CEO Dara Khosrowshahi sent an email announcing the company would begin treating hiring as a “ privilege ” in response to “a seismic shift” in the markets. Slowing hiring is one thing, but tech companies also seem to be increasingly open to outsourcing employment to China and India. The increased prevalence of remote work, steep H-1B visa restrictions and the shortage of skilled U.S. workers are all driving this trend, according to the Wall Street Journal. The optics of outsourcing are never popular domestically, but that could be particularly true in an economic recession that puts more Americans in precarious employment situations. Republicans are also primed for a sweeping victory in midterms, and the Trump wing of the party has placed a strong emphasis on domestic hiring and cracking down on H-1B workarounds. The crypto lobby has a harder case to make Six months ago, it would be much easier to argue bitcoin should be allowed in 401(k)s. With bitcoin losing around half its value between then and now, it seems like an obviously bad idea to put your retirement funds in a speculative digital asset. This isn’t a hypothetical debate — in April, Fidelity announced it would allow users to invest as much as 20% of their retirement savings in bitcoin, a decision that elicited condemnation from Sens. Elizabeth Warren and Tina Smith. More broadly, crypto skeptics stand on firmer ground now that NFT and crypto markets have been battered. That puts a higher burden of proof on the industry and its lobbying group at a time when the regulatory environment is particularly malleable. New York, Texas and Florida have all been racing to set the tone for national crypto legislation. This dynamic is even clearer once you consider specific issues. For example, the crypto lobby has been pushing New York to ease restrictions on BitLicense requirements, which limit the types of digital assets that centralized exchanges can offer. The lobby would have a much easier time arguing in favor of eased restrictions with asset prices up. But with many overzealous investors facing financial ruin from crypto losses , restrictions on asset types look like a responsible and perhaps even necessary guardrail. Put the oversized checks in storage Lobbying the government has reliably produced some of the world’s highest “investment” returns. The semiconductor sector, for instance, will likely see a 500x return on a $100 million lobbying budget. Politicians have grown accustomed to touting these sweetheart deals as win-wins. Tech companies were handed hundreds of millions of dollars so that politicians could hold press conferences where they say things like , “For every six cents of capital investment Ohio will make, Intel will make a dollar.” Of course, these deals don’t always go over well. In Georgia, for instance, the decision to lure EV maker Rivian with subsidies has become a highly contentious issue . If the recession hits hard, the subsidies-for-expanded-tax-base model will likely become less popular. The headlines practically write themselves: “Trillion-dollar company receives millions in taxpayer subsidies while everyone else suffers.” We’ve seen this narrative play out with the anti-bailout movement that came out of the 2008 financial crisis. It’s difficult to say whether this dynamic will translate into reduced subsidies — it could just mean fewer oversized-check photo ops at press conferences. Keep ReadingShow less