Tech companies want humans to help level up AI models. What's your price for training them?
Happy almost Friday! It's not TV. It's HBO, HBO Go, HBO Now, HBO Max, Max, HBO Max. The TV brand that brought us "The Sopranos" and "Sex and The City" is rebranding its streamer … again. BI's Peter Kafka has more on all the flip-flopping and why, jokes aside, it's acknowledgment of a deal gone bad. (At least HBO is leaning into the joke.)
In today's big story, Meta will pay you $50 an hour to help make its avatars more realistic. Which begs the bigger question: How much would you need to be paid to train AI on something you're good at?
What's on deck
Markets: The US-China trade deal doesn't mean investors are out of the woods yet.
Business: The job market might be the biggest risk no one's talking about.
But first, lending a helping hand to AI.
If this was forwarded to you, sign up here.
The big story
Train and gain
Tayfun Coskun/Anadolu Agency via Getty Images
Want to make a quick $50? Just say "Cheese."
Meta is paying $50 an hour to record people's facial expressions and small talk to bolster its virtual avatars, writes BI's Effie Webb.
It's part of the tech giant's massive bet that the metaverse is the future. Meta's recruiting adults to make its virtual reality avatars as realistic as possible and create a social environment that is "indistinguishable from reality" during virtual interactions.
That sounds exhilarating or terrifying, depending on where you sit on the technology adoption scale. What's not up for debate is that plenty of this type of work is going around.
Across the tech world, companies are hiring humans to train AI models to act more … human. Tesla previously paid up to $48 an hour to people training its humanoid robots. Meta also has contractors monitoring how "flirty" its chatbots get.
The humanization of AI is turning into a nice side hustle for those willing to put in the hours. One woman told BI she made $8,000 in three weeks training AI after hours. But it's not always easy work. Sometimes it can involve stress-testing AI with "harmful" prompts, as Effie previously reported.
Sharing your smile is one thing, since most of us don't get paid to flash our pearly whites. (Don't worry. I still think you look great.)
But what about training AI on something you're good enough to get paid to do?
On the one hand, you could probably demand a higher rate for teaching AI a somewhat proprietary skill. On the other hand, are the short-term gains worth the long-term risks that could come with AI eventually being better at the task than you?
Simply put: Do you want to be the turkey setting the table for Thanksgiving?
Let me know what you think. They say everyone has a price. What's yours for teaching AI your secret sauce?
3 things in markets
Reuters / Lucas Jackson
1. The S&P 500 just flashed an optimistic technical signal. The index moved above its 200-day moving average, which has historically signaled a positive long-term trend. LPL Financial's technical analyst said it could spell a bottom for the market.
2. And Wall Street's fear is subsiding. The index known as Wall Street's fear gauge saw its fastest-ever drop from April 10 to May 12; Wall Street's breathing easy now that the US and China have dialed down tariffs. The index had peaked when fears were high right after Liberation Day.
3. But investors might be declaring tariff victory too early. The US-China deal to temporarily reduce tariffs on Chinese imports to 30% pushed the S&P 500 up 3.2%. However, investors might be getting ahead of themselves for a variety of reasons.
3 things in tech
1. Tariffs have taken a toll on TikTok's US e-commerce business. Four TikTok staffers told BI they attribute the drop-off in US sales to tariffs, which have created cost headaches for merchants. It's one of the many setbacks the social media platform has experienced this year.
2. Waymo just got another recall headache. The company issued a software recall for more than 1,200 robotaxis after several of its cars were involved in barrier collisions, according to a report filed with a federal safety agency. It's Waymo's third recall in over a year, though the last two were voluntary.
3. Amazon robots won't take your job, but they might end the need for it. Amazon said it sees warehouse robots as crucial to "flattening" its hiring curve, according to an internal document obtained by BI. It suggests Amazon could use robots to slow the rate of new hiring rather than replace existing workers.
3 things in business
1. The job market is on the brink of collapse. While unemployment remains historically low, things may not be as tranquil as they seem, writes economist Neil Dutta. From fewer job openings to the uncertainty surrounding Trump's trade upheaval, signs are pointing to higher unemployment and a weaker job market.
2. Behind Ryan Coogler's deal to own "Sinners." The director behind the box-office sensation struck a rare deal with Warner Bros., giving him the rights to "Sinners" in 2050. A veteran entertainment and technology attorney spoke to BI about how Coogler could cash in.
3. The key deals struck during Trump's tour of Saudi Arabia. Companies like Nvidia and Amazon have notched some big wins while Trump has been in the Middle East. Here's a list of what's been announced, from a major chip deal to a planned Saudi AI zone.
In other news
Trump can't stop talking about the pushback to Qatar's Air Force One offer.
We asked Grok why it was bringing up 'white genocide' in unrelated X posts. The AI's answers are messy.
Uber has a new plan to get you to use its app more: Cheaper ride and delivery options.
What's happening today
The Business Insider Today team: Dan DeFrancesco, deputy editor and anchor, in New York. Hallam Bullock, senior editor, in London. Grace Lett, editor, in Chicago. Amanda Yen, associate editor, in New York. Lisa Ryan, executive editor, in New York. Ella Hopkins, associate editor, in London. Elizabeth Casolo, fellow, in Chicago.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
31 minutes ago
- Yahoo
AppLovin: The advertising giant in the making
Introduction AppLovin is a premier mobile technology company that has transitioned from a mobile gaming developer to a full-fledged app marketing and monetization platform. It was established in 2012 and is based in Palo Alto, California. The company operates two main business units: Software Platform and Apps. The Software Platform, which is driven by the AI-based AXON engine, is aimed at mobile app developers and it provides them with cutting-edge solutions like user acquisition, monetization, and analytics. This platform is responsible for handling billions of ad requests on a daily basis, as it leverages machine learning algorithms to improve campaign performance and to increase revenue for both advertisers and publishers. Among the key products offered are the MAX mediation platform which is the instrument that allows publishers to operate several ad networks with a single integration, and thorough user acquisition tools that are specifically designed for developers to grow their user bases at the shortest time possible. Warning! GuruFocus has detected 3 Warning Sign with APP. Key Growth Drivers Machine Learning and AI-Powered Optimization The AXON platform that is powered by AppLovin's sophisticated machine learning infrastructure, which is its own product, is the genetic code of the tech advantage of the company. The proprietary AI engine that the company uses, being at the heart of the whole process, is in charge of the incredible computational task of processing a diverse range of data to make real-time decisions on the best ad placements, user acquisition campaigns, and monetization strategies available. By examining the engagement metrics and conversion data, the system constantly improves the targeting accuracy and campaign efficiency besides getting insight from user behavior patterns. This exceptional efficiency, which allows advertisers to make more informed bids on ad inventory while publishers to generate more revenue, is based on the system's predictive capabilities. The platform's establishment to process and learn from billions of pieces of data each day creates a significant barrier for other competitors, as it is developed better with more available data. AXON is the neural network that drives the whole platform with deep learning technology, which is used to explore the user data in a way that traditional analysis tools would not be able to. That is to say, this platform is the most modern digital solution that performs the task of reinforcement learning, where the smart algorithms are given real-time feedback with optimal ad placement and bidding decisions generated automatically. This automated optimization function, which in reality results in the use of less human resources, is achieved compared to the delivering of impressive results using a set of predefined rules. Real-Time Bidding and Programmatic Infrastructure AppLovin has the technology stack including high-performance real-time bidding (RTB) infrastructure capable of processing hundreds of thousands of bid requests per second with sub-100 millisecond response times. This is a low-latency system for mobile advertising where split-second decisions are the determinants of campaign success. The distributed architecture of the platform ensures global scalability and uniform performance, despite different geographic regions. Moreover, the programmatic advertising technology platform includes header bidding alongside the integration allowing publishers to increase their revenue through enabling multi-demand sources to compete for the same inventory simultaneously. The system's complex auction mechanics are not only concerned with bid prices but also take into account user experience, ad quality, and long-term value optimization. However, Applovin's server-side integration mechanisms spare the technical brunt of application developers while presenting centralized data collection and analysis techniques. This approach is compared to client-side implementations, which need multiple SDK integrations, resulting in longer app latency and poor user experience. Data Analytics and Attribution Technology The advanced data analytics structure of the company processes the huge amounts of data it collects to find insights that are valuable for both the advertisers and publishers. More advanced attribution models, in turn, allow for better measurement of the user paths across multiple touchpoints, which translates into higher confidence in the budget allocation decisions. The two ways to attribute the platform's capability are probabilistic and deterministic, which are useful for a much dense privacy environment. AppLovin's analytics tool provides sophisticated analysis of cohorts, predicting lifetime value, and churn modeling to the app developers in an effort to enhance the user acquisition process. The platform's ability to connect the costs of acquiring new users with the value they bring over time means it can help optimize return on ad investment (ROAS) more than measuring just installs can. The company's commitment to using privacy-friendly methods in data collection and analysis ensures that it is well-prepared for the ever-changing legal landscape. Data confidentiality and federated learning methods, together with innovative approaches in data collection, will facilitate users' right to privacy, and continue the optimization process by respecting regulatory mandates. Competition from Technology Giants AppLovin is up against robust competition from technology companies that are using their resources and networks. Google, through its AdMob platform and Google Ads ecosystem, is probably the strongest rival. Google's interconnectedness with Android, YouTube, and its advertising network turned it into a powerhouse that promises premium exposure and fine contracts. Facebook (Meta), through its Audience Network along with the vast amount of data from social media, which makes it possible for precise user targeting and cross-platform campaign management, is another relevant competitor. Apple's recent privacy policies, namely, App Tracking Transparency (ATT), are a mixed-source benefit in the competition scene. The fact is that these issues eliminate some of the attribution and targeting capabilities for a while but at the same time, that would bring balance to the situation by lessening the data advantages that some of the competitors had before. The way to deal with this threat is a concentration on those areas where the giants might not be as swift or/and comprehensive. The deep dive on the mobile gaming and app monetization areas by the company grants the right for more in-depth verticals and exactly tailor-made solutions rather than platform companies do. The optimization brought along with the AXON platform's automatic algorithms for the mobile app environment often results in better performance than generic ad solvers in the scope of new user quality and monetization efficiency. Competition from Specialist Ad Tech Companies The mobile advertising game is competitive with the participation of many niche players, each of whom has identified a particular slice of the market. Unity Technologies is one of them, in particular, because it uses its Unity Ads platform to operate in the same arena as mobile gaming ads and even leverage its game development engine to provide integrated advertising. IronSource (now part of Unity) used to be a significant adversary with its all-in-one platform that handled app monetization and user acquisition comprehensively. Chartboost has a very specific operation model that enables it to focus only on mobile game advertising, which includes direct publisher relationships and extensive knowledge of the gaming industry. Vungle (now part of Liftoff) is a firm that, through its video ad solutions, has added strong creative optimization capabilities. These companies are mostly known for their vibrant industry knowledge and long-standing partnerships defined within the industry sectors. However, these specialized players are giving a tough time with their niche focus, AppLovin is competing against them by bringing extra scale and high-level technology. The yield of billions and billions of ad requests daily makes the company to have the capacity of more effective machine learning optimizationin and standard machine learning schemes. AppLovin's user acquisition, monetization, and analytics in one solution platform and the simplicity of not having to carry out multiple integrations delivers accessibility and effectiveness over single point solutions that require integrations. Strengthened by the right acquisitions, the company is in a stronger position as it has consolidated its market share, and the potential competition was eliminated. The MoPub acquisition from Twitter gave AppLovin a digital publisher relationship and mediation capabilities while other small acquisitions enriched the tech stack and human resources, thus enhancing the platform. Key Risks Risks of Platform Dependency and Ecosystem Control AppLovin is at a high risk concerning the technological factors stemming from its reliance on mobile platforms governed by Apple and Google. Any iOS and Android platform policy changes can, in a matter of seconds, thus alter the way AppLovin operates. The real-time impact of the implementation of Apple's App Tracking Transparency (ATT) framework proved this vulnerability when mobile advertisers immediately lost the ability to accurately attribute and target their ads. The prospective platform issues may involve more rigorous SDK approval processes, more privacy restrictions, or modified app store policies that will limit the advertising functionality altogether. For instance, the introduction of Private Relay in iOS 17 and the addition of enhanced tracking protection greatly narrow data collection capabilities. Although the Privacy Sandbox initiative from Android is still in progress, it poses a serious risk of redefining the way mobile advertising attribution and targeting are done. The technical risk branch out to the potential scenarios of platform fragmentation in which different versions of iOS or Android impose different privacy and advertising restrictions. AppLovin's single platform strategy is rendered more complex as it needs to reconcile with the different technical requirements caused by the braid of platform versions and geographic regions with various regulatory frameworks. Algorithmic and Machine Learning Model Vulnerabilities AXON's machine learning algorithms are the backbone of AppLovin's business edge as they grant a strongly concentrated technical risk around model performance and accuracy. The susceptibility of machine learning models to concept drift, where changing user behavior patterns or market conditions reduce the effectiveness of the model over time, is one of the most common reasons for this fall. The constant mutation of the mobile advertising ecosystem can make the previously useful alphas to be of no use whatsoever. Model bias is another highlight technical risk, this is when the algorithms inadvertently discriminate against some user segments or app categories, which in turn could result in regulatory violations or suboptimal performance outcomes. Attacks that are deceitful to the machine learning systems could influence either the bidding algorithms or the attribution models, which will in turn lead to financial losses and site degradation. The scenario of being data poisoned where the hackers introduce the useless and malicious data into the training datasets has the detrimental effect of degrading the performance of the models or creating vulnerabilities which can be exploited. The scale at which AppLovin is processing data, daily handling over billions of events, is making it more difficult to ensure comprehensive data validation while maintaining the real-time processing needs. The most serious of the risks that are created through overfitting is the one that comes from the models completely specializing in what has been done historically, thereby rendering the generalization capacities for new market conditions or user behaviors very weak. The high focus on the gaming vertical that has led to a competitive advantage could be detrimental if the gaming industry undergoes a drastic change. Valuation The forward P/E ratios for AppLovin portray impressive earnings acceleration with the non-GAAP P/E declining from 45.93x (FY1) to 27.69x (FY3), which suggests an impressive earnings increase, outstripping all but one peer company. It is the market's strong response that makes the compression evident, showing that it appreciates AppLovin's ability to grow the business while also keeping profitability up. The company's PEG ratios of 0.94 (non-GAAP forward) and 0.31 (GAAP TTM) are at a high level with respect to the growth rates making the stock very attractive, at least, it is significantly more attractive than peers like Adobe (1.35) and Cadence (3.09). This fact shows that the value of AppLovin is off the charts due to its markdown price based on its growth potential, which is further elaborated by the large upside that is now available. In contrast to industry rivals, AppLovin's valuation metrics look realistic even if the absolute P/E ratios are higher. Adobe and Cadence declare lesser P/E compression over time, while AppLovin shows signs of higher earnings growth. The absence of profit for a company like MasterCard (negative P/Es) contrasts sharply with AppLovin's prevailing profits in the ad-tech sector. Guru Holdings Lowenstein's 17.19% stake which is equivalent to $762.85 million shows tremendous conviction, especially if we take into consideration the average buy price that he had of $75.06, which is representing a 423.6% gain. Lowenstein's convincing position, which is large in size and yields excellent returns, is an evidence of AppLovin's strategic execution and its growth path. The 12.86% increase in the holdings that Lowenstein took just lately proves that he still has confidence in the company despite the stock's larger rise, which in turn shows that the bottom line is the company's fundamentals rather than the ups and downs of the market. Resnick's 13.47% stake ($740.26M) with an average cost basis of $49.41 (695.4% gain) represents even earlier conviction in AppLovin's transformation story. The stability of his holdings (0% recent change) indicates dissatisfaction with current positioning while maintaining long-term conviction. Both managers' five-star ratings and substantial outperformance demonstrate their investment expertise. ConclusionAppLovin faces strong competition from technology companies like Google, Facebook, and Apple. Google's AdMob platform and Ads ecosystem, coupled with its interconnectedness with Android, YouTube, and its advertising network, offer premium exposure and fine contracts. Facebook's Audience Network and vast data from social media enable precise user targeting and cross-platform campaign management. Apple's recent privacy policies, App Tracking Transparency (ATT), provide mixed-source benefits in the competition scene, but may limit data advantages. To address this threat, AppLovin focuses on mobile gaming and app monetization areas, offering tailored solutions rather than generic ad solvers. The AXON platform's automatic algorithms for the mobile app environment often result in better performance than generic ad solvers in terms of user quality and monetization efficiency. This article first appeared on GuruFocus. Sign in to access your portfolio
CNET
36 minutes ago
- CNET
Protect Your Property With the Low Maintenance Eufy SoloCam S220 for Just $65
According to a recent CNET survey, one in six US adults has been a victim of package theft at least once -- and that's just one of many growing threats. This is exactly why having a reliable outdoor security camera isn't just a nice-to-have: It's essential for added protection and a little peace of mind. Plus, if you've been thinking about getting one, now's the perfect time to make it happen. Eufy, one of our go-to home security brands, is offering a huge 50% off its solar-powered S220 security camera, bringing the price down to just $65. There are no monthly fees or hidden costs so it's a one-and-done purchase. The only catch? This promotion is part of a limited-time deal, so you might want to act fast. This tiny but mighty camera comes in a wire-free design and takes approximately five minutes to set up. It features an IP67 rating, so whether it's rain, shine or snow, it's built to handle whatever the weather throws at it. Hey, did you know? CNET Deals texts are free, easy and save you money. You'll get a clear 2K video that shows exactly what's happening outside your home. The f/1.6 night vision gives you a sharp view even in the dark, and the built-in AI can tell the difference between people and random objects -- you won't be bombarded with false alarms that just stress you out. You can even set custom security zones to get alerts only when there's motion where it actually matters. The Eufy S220 also works with voice assistants like Alexa and Google, so you can control it completely hands-free. And with two-way audio, you can easily communicate with whoever's at your door without having to get up. Finally, you don't have to worry about recharging the camera at all. Just 3 hours of sunlight a day keeps the solar battery up and running. Why this deal matters Taking effective measures to protect your property is no longer optional. This solar-powered security camera is not just easy to maintain but also offers a bunch of advanced features, now for just $65. This is one of the best prices we have seen on the S220, but the discount might not last long, so take advantage of it while you can.
Yahoo
37 minutes ago
- Yahoo
Wells Fargo Upgrades Zscaler (ZS) Stock, Lifts PT
Zscaler, Inc. (NASDAQ:ZS) is one of the 10 software stocks analysts are upgrading. On June 13, Wells Fargo upgraded the company's stock to 'Overweight' from 'Equal Weight,' lifting the price objective to $385 from the previous target of $260. As per the firm, software infrastructure spending in H2 2025 is expected to be driven by the adoption of agentic AI, which tends to benefit several software companies, including Zscaler, Inc. (NASDAQ:ZS). An employee standing in front of a large data center, looking toward the future of cloud security. The research firm cited Zscaler, Inc. (NASDAQ:ZS)'s potential to reach $5 billion in annual recurring revenue by fiscal 2027 as one of the measures of the upgrade. Despite the increase in the stock price YTD, the firm believes it still possesses significant room for multiple expansion. Well Fargo expects more than 20% billings growth in FY 2026 and continues to see margin expansion continuing, together with revenue growth. Notably, the company's stock has increased by ~66% on a YTD basis. In May 2025, Zscaler, Inc. (NASDAQ:ZS) signed a definitive agreement to acquire Red Canary, which is a leading managed detection and response (MDR) vendor. Through the combination of Zscaler, Inc. (NASDAQ:ZS)'s high-volume and high-quality data with Red Canary's domain expertise in MDR, Zscaler, Inc. (NASDAQ:ZS) plans to ramp up its vision to deliver AI-powered security operations. While we acknowledge the potential of ZS to grow, our conviction lies in the belief that some AI stocks hold greater promise for delivering higher returns and have limited downside risk. If you are looking for an AI stock that is more promising than ZS and that has 100x upside potential, check out our report about this cheapest AI stock. READ NEXT: 13 Cheap AI Stocks to Buy According to Analysts and 11 Unstoppable Growth Stocks to Invest in Now Disclosure: None. Sign in to access your portfolio
