Amazon has introduced an upgrade to Alexa+, its next-gen AI assistant, enabling users to order food from popular delivery platforms like Uber Eats and Grubhub in a conversational style, much like interacting with a server at a restaurant or making an order at a drive-thru, as stated by Amazon.

This latest Alexa+ functionality aims to foster a more intuitive and engaging ordering process. Beginning today, users can request a type of cuisine, browse menu selections, pose questions, and personalize their meals all in a single dialogue. Should users wish to modify their order midway, add a dessert, or change quantities, they can do so seamlessly. 

To utilize this feature, users must connect their Grubhub or Uber Eats accounts via the Alexa app. After linking, past orders will sync automatically, simplifying the process of reordering favorite dishes or exploring new eateries. Users can simply say, “I’d like to order Italian for delivery,” and Alexa+ will assist by presenting various restaurant choices. Once the order is finalized, users receive a detailed summary of the items in their cart, including amounts and prices. 

This innovative food ordering functionality is currently being rolled out to Alexa+ users with Echo Show 8 devices and beyond. Amazon articulates that this step represents a notable enhancement in the company’s objective to develop adaptive interaction models. This effort sets the stage for the potential expansion of similar features into other sectors, like grocery shopping and travel planning.

The Alexa+ enhancement arrives amidst the broader consequences of AI within the food sector. Fast food establishments have already begun employing AI assistants in drive-thrus, but challenges persist, particularly regarding order precision. In 2024, McDonald’s halted its initiative following a few errors, including an AI cashier inadvertently adding nine sweet teas to a customer’s order. Furthermore, Taco Bell encountered its own set of issues, with viral footage highlighting its AI making mistakes.

Since the launch of Alexa+ in the U.S. and its recent extension to the U.K., the features continue to evolve, with the introduction of new personality modes for Alexa, such as a “Sassy” option aimed at adults, alongside other styles including Brief, Chill, and Sweet. 

Whoop, the wearable tech company focused on fitness and health monitoring, has successfully completed a $575 million Series G funding round at a valuation of $10.1 billion — nearly three times its previously reported valuation of $3.6 billion — in a deal involving sovereign wealth funds, leading health organizations, and some of the globe’s most famous athletes.

The funding round was spearheaded by Collaborative Fund and saw participation from Mubadala Investment Company, Qatar Investment Authority, 2PointZero Group, Abbott, Mayo Clinic, Macquarie Capital, IVP, Foundry Group, Accomplice, Affinity Partners, Glade Brook, B-Flexion, Promus Ventures, and Bullhound Capital. Notable individual investors include Cristiano Ronaldo, LeBron James, Rory McIlroy, Reggie Miller, and Niall Horan, alongside various other high-profile athletes and celebrities.

The company has accumulated around $900 million in total funding since its inception.

A significant player joining the cap table is Abbott, the prominent medical device company. Whoop’s founder and CEO Will Ahmed indicated that this partnership represents a larger initiative aimed at enhancing health and medical capabilities, though he mentioned there is “more to come” regarding that particular announcement.

This funding is timely as Whoop reaches several important business milestones, according to Ahmed. The company closed last year with a bookings run rate of $1.1 billion, reflecting a 103% increase year over year. In a conversation with TechCrunch last week, Ahmed emphasized why bookings should be the metric of focus: When you are distributing millions of hardware units globally while managing a subscription model, investors must comprehend the financial dynamics of simultaneously handling inventory, hardware expenses, and recurring revenue. It presents a more intricate scenario than a traditional software company, and bookings effectively encapsulate that complexity.

Regarding the future utilization of this capital, Ahmed mentioned priorities like talent acquisition and recruitment, elevating marketing and brand recognition, ongoing R&D investments, alongside expediting international growth.

The pressing question surrounding such a substantial funding round at this valuation: Is an IPO on the horizon? (Competing firm Oura is reportedly discussing plans with bankers for its own IPO this year.) Ahmed stated that the company is engaged in “a lot of the no-regrets work to be a public company,” but refrained from indicating any immediate intentions to go public.

Considering that Whoop is a consumer brand with significant recognition among health-focused and performance-driven individuals, it’s the sort of enterprise that could spark genuine enthusiasm among retail investors whenever it opts to take that step. Meanwhile, Whoop has plenty of runway ahead and a much larger figure next to its name.

You can tune in to our comprehensive dialogue with Ahmed, where we also discussed the company’s foundational days, its extensive hiring strategy at present, and how Whoop is integrating AI into its operations. Alternatively, you may read about Whoop’s ambitious foray into healthcare and its implications for the brand here.

Uber and the Chinese autonomous vehicle firm WeRide have initiated robotaxi services without a human safety operator in Dubai, marking a significant expansion in the Middle East.

Users can now request these vehicles via Uber’s app, with services available in commercial and industrial zones such as Dubai Silicon Oasis, Dubai Investment Park Second, and Jabal Ali Industrial First, as well as suburban regions and the maritime trade center Al Hamriya Port. The service is managed locally by Tawasul, a fleet and mobility operator in the UAE.

The firms initially presented robotaxis in Dubai last December under a trial scheme, offering free rides and utilizing a human safety operator. Last month, the Roads and Transport Authority of the government issued a permit for the companies to trial driverless vehicles.

This launch indicates a significant advancement in WeRide’s robotaxi initiatives in the area and highlights Uber’s ongoing partnership with the firm. According to filings submitted to the U.S. Securities and Exchange Commission on Monday, Uber possesses a 5.82% interest in WeRide. Last year, Uber invested $100 million in WeRide, which places Uber’s equity stake at approximately $150 million based on WeRide’s stock closing price on Monday.

“Introducing fully autonomous vehicles to Dubai represents a critical step toward achieving global autonomous mobility,” stated Sarfraz Maredia, Uber’s global head of autonomous mobility and delivery. “This launch epitomizes our strong commitment to the UAE and our vision for a hybrid ecosystem—where drivers alongside AVs cultivate a more robust network. Particularly during trying times in the region, we take pride in being a partner to this city, ensuring that Uber remains a reliable option for seamless and confident transport.”

In May 2025, Uber expanded its investment in WeRide as part of a commercial robotaxi collaboration aimed at extending service to another 15 cities in the next five years. The companies have previously indicated that this expansion will encompass cities in Europe.

Through this partnership, WeRide’s robotaxi offerings can be accessed via the Uber app. This agreement is akin to Uber’s arrangement with Waymo, where Uber manages network routing and fleet operations, while the autonomous vehicle firm oversees the AV technology.

Correction March 31 at 12:15 p.m. PT: A previous version of this article mistakenly mentioned that Uber had increased its stake in WeRide. It also inaccurately stated the value of Uber’s equity stake in WeRide; it is approximately $150 million as of Monday.

A suspected hacker from North Korea has taken control of a widely used open source software development tool, modifying it to deploy malware that could endanger millions of developers.

On Monday, a hacker released malicious versions of the popular JavaScript library known as Axios, which developers depend on to enable their software’s internet connectivity. The compromised library was made available on npm, a repository for open source project code. Axios is downloaded by users tens of millions of times each week. 

The hijacking was detected and halted within approximately three hours from Monday night into Tuesday, as reported by the security firm StepSecurity, which investigated the incident. 

Cybercriminals are increasingly targeting developers of prominent open source projects to execute mass hacks against anyone relying on the compromised code, thereby potentially acquiring access to numerous affected devices. Such widespread breaches are labeled supply chain attacks because they exploit software that enables hackers to then compromise anyone who downloaded the tainted software. In recent times, attackers have focused on firms such as 3CX, Kaseya, and SolarWinds, in addition to open source tools like Log4j and Polyfill.io, to target extensive user bases.

Currently, it remains uncertain how many individuals downloaded the malicious variant of Axios within that timeframe. The security firm Aikido, which also looked into the incident, advised anyone who obtained the code to “assume their system is compromised.”

Google informed TechCrunch that its security analysts are associating the Axios breach with North Korean hackers.

“We have attributed the attack to a suspected North Korean threat actor we track as UNC1069,” stated John Hultquist, the chief analyst for Google’s Threat Intelligence Group. “North Korean hackers possess significant experience with supply chain attacks, which they have historically employed to steal cryptocurrency. The complete scope of this incident remains unclear, but due to the popularity of the compromised package, we anticipate it will have widespread consequences.”

The hacker managed to insert malicious code into Axios by breaching the account of one of the project’s main developers, who had the authority to release updates. The hacker replaced the legitimate developer’s email address on the account with their own, complicating the process for the developer to regain access.

After taking control of the account, the hacker added harmful code intended to deliver a remote access trojan, or RAT — essentially malware that allows hackers complete, remote control over a victim’s machine. The hacker subsequently issued new versions of Axios disguised as a legitimate update for Windows, macOS, and Linux users. 

The hackers also programmed the malware, along with some of the code used for its delivery, to automatically erase itself post-installation in an effort to evade detection by anti-malware software and investigators, according to security experts.

Updated to include information from Google regarding the attribution to North Korea.

Warehouses are rapidly embracing automation. While firms like Amazon are creating their own robotic fleets internally, others are seeking automation technology from external providers.

FedEx has explored both pathways and determined that collaborating with robotics firms is the optimal strategy to stay competitive in the automation arena.

FedEx’s latest multi-year alliance with Berkshire Grey, a robotics company owned by SoftBank, exemplifies this strategy: leveraging expert knowledge to design robots capable of performing repetitive, hazardous tasks traditionally handled by humans. This non-exclusive partnership has led to the creation of Scoop, a robot intended for unloading bulk packages, allowing for the removal of large quantities of parcels from trucks simultaneously.

FedEx plans to introduce these robots in its warehouses via a pilot initiative later this year. While these robots won’t service every one of FedEx’s many unloading doors, the company aims to scale the bot’s deployment if the pilot is successful.

Stephanie Cook, director of advanced technology and innovation in robotics at FedEx, shared with TechCrunch that bulk unloading ranks among the most physically demanding and unpredictable tasks within FedEx’s warehouses. This isn’t the company’s first endeavor to automate bulk unloading; Cook mentioned challenges in identifying the appropriate robot for this function.

“We haven’t found anything off-the-shelf that meets our needs,” Cook stated. “We collaborated with Berkshire Grey previously and determined it was a suitable partnership for us. We recognized that this wasn’t a project that could be developed within a few months; it was going to be a multi-year journey to achieve this.”

According to O.P. Skaaksrud, vice president of advanced technology and innovation at FedEx, bulk unloading is an ideal function for automation. Although it does require these robots to make decisions, the complexity is lower than selecting or searching for specific packages, making the process easier to automate.

“Given the diversity of package types, focusing on individual picks is simply not fast enough,” Skaaksrud explained. “That was one of the trade-offs we considered, as there are package unloaders that do single picks, but they lack the speed to handle this variety.”

Cook noted that the company is prioritizing the automation of the most hazardous and physically strenuous roles within its warehouses. These jobs are better suited for automation in general since they are often repetitive, freeing up employees to engage in safer, more skilled work.

Picking and packing partners

The Memphis-based firm does engage in in-house technology development, such as the FedEx SenseAware and SenseAware ID sensor systems, which facilitate package tracking.

However, Skaaksrud emphasized that creating sensors and developing robotics are distinct challenges.

“The development of sensor hardware for packages is complex, but advancing robotics capabilities is of an entirely different caliber,” Skaaksrud remarked. “Partnering with other companies in the field accelerates our progress. That’s how we perceive these collaborations — very advantageous for both FedEx and our partners.”

Berkshire Grey isn’t the only automation-centric partnership for FedEx. The company has secured several alliances (and conducted trials) in recent years as it aims to automate more aspects of its operations both inside and outside the warehouse.

Inside the warehouse, FedEx partners with Dexterity, a robotics startup valued at over a billion dollars that specializes in robots engineered to mimic human actions. Additionally, they have an agreement with another unicorn startup, Nimble, which focuses on creating fully autonomous warehouses.

Autonomous deliveries, whether for last-mile or long-haul, are also a key area of interest.

In 2021, the company entered into a pilot agreement with Aurora Innovation, an autonomous trucking startup, to transport packages along defined routes in Texas with self-driving trucks. The collaboration was expanded in 2022, resulting in over 3,200 successful autonomous deliveries.

FedEx also formed a partnership with Nuro, a company specializing in autonomous last-mile delivery robots, in 2021. At that time, FedEx characterized it as a long-term commitment with intentions to expand last-mile delivery capabilities. However, Nuro shifted from delivery to licensing its autonomous technology in 2025, leading to the conclusion of their collaboration.

The last-mile delivery avenue has been one that FedEx has explored internally with mixed results. The company developed and launched the SameDay Bot in 2019 to aid with last-mile delivery, but the bots faced criticism and were eventually banned from NYC by former mayor Bill de Blasio. After a few years, FedEx moved away from these efforts while reaffirming that it remains an area of focus.

Keeping it real

While the company is keen on remaining competitive in the overarching automation landscape, both Skaaksrud and Cook stressed that FedEx will not rush ahead impulsively. They are not simply chasing the next flashy robot or technology as soon as it becomes available.

“If we concentrate solely on the technology, we’ll likely fail,” Skaaksrud cautioned. “This is akin to playing 3D chess; you must address numerous components, many of which may not seem exciting, that are crucial to the overall solution. We are certainly taking steps to not only have compelling technology but also productive technology that effectively addresses business challenges.”

The company is also unconcerned that its partnership-based strategy will lead to a lack of proprietary technology. Skaaksrud posited that the hardware is merely hardware, asserting that FedEx trucks are simply trucks, and it’s the network and intelligence behind the delivery system that defines FedEx.

While headlines might suggest that every company is fervently pursuing full automation, FedEx intends to remain judicious as it introduces new technologies.

For Cook, the primary emphasis of these strategies is still on the human workforce within the warehouses, meaning technology must be crafted to complement their roles, enhancing safety and ease of work.

Consequently, and given the company’s priority on areas with clear returns on investment, humanoid robots are not a focus.

“Managing multiple humanoids in a constrained, dynamic space is quite challenging,” Skaaksrud noted. “I find humanoids to be fascinating and we’re definitely monitoring advancements, but it’s about finding the right application. You must navigate the hype as expectations can be misleading, yet there remains substantial long-term potential. Understanding limitations is crucial to setting realistic expectations.”

To create the autonomous machines of tomorrow, sometimes a model is required for your model. 

Businesses that are working on self-driving vehicles, robots interacting with the physical world, or autonomous construction machinery gather thousands, if not millions, of hours of video footage for analysis and training. 

Sorting and archiving that video is currently a task for humans, who must view all of it. Even when fast-forwarding, that process doesn’t scale. NomadicML, a company established by CEO Mustafa Bal and CTO Varun Krishnan, aims to address issues for clients who have 95% of their fleet data stored away.

The difficulty increases when searching for edge cases — the most crucial data showcases events that seldom happen and can confuse inexperienced physical AI models.

Nomadic is aiming to tackle that issue with a platform that converts footage into a structured, searchable dataset through a suite of vision language models. This, in turn, enables improved fleet monitoring and the crafting of distinctive datasets for reinforcement learning and quicker iteration.

The company revealed an $8.4 million seed funding round on Tuesday at a post-money valuation of $50 million. This round was led by TQ Ventures, with contributions from Pear VC and Jeff Dean, and will facilitate the onboarding of more clients and the ongoing enhancement of its platform. Nomadic also secured first place in Nvidia GTC’s pitch contest last month. 

The two founders, who met as undergraduates in computer science at Harvard, “kept encountering the same technical obstacles repeatedly at our jobs” at firms such as Lyft and Snowflake, Bal shared with TechCrunch. 

“We offer individuals insights into their own footage, whatever drives their own AVs [and] robots,” he stated. “That is what propels these developers of autonomous systems forward, not arbitrary data.”

Consider, for instance, trying to enhance an AV’s comprehension that it can proceed through a red light if a police officer commands it to do so, or identifying every instance when vehicles pass under a particular type of bridge. Nomadic’s platform enables these occurrences to be recognized for both compliance reasons and to be directly integrated into training pipelines. 

Clients such as Zoox, Mitsubishi Electric, Natix Network, and Zendar are already leveraging the platform to create intelligent machines. Antonio Puglielli, the VP of Engineering at Zendar, remarked that Nomadic’s tool enabled the company to accelerate its work much more swiftly than the option of outsourcing, and that its domain expertise distinguished it from other competitors.

This type of model-based, auto-annotation technology is emerging as a vital workflow for physical AI. Established data labeling companies like Scale, Kognic, and Encord are creating AI tools to conduct this work, while Nvidia has launched a series of open-source models, Alpamayo, that can be tailored to address the issue.

Varun contends that his company’s tool goes beyond mere labeling; it functions as an “agentic reasoning system: you specify what it needs, and it determines how to find it,” utilizing multiple models to comprehend actions occurring and contextualize them. Nomadic’s investors anticipate that the startup’s focus on this specific infrastructure will prevail. 

“It’s akin to why Salesforce doesn’t create its own cloud and Netflix doesn’t construct its own [content distribution facilities],” Schuster Tanger, a partner at TQ Ventures who spearheaded the round, told TechCrunch. “The moment an autonomous vehicle company attempts to internally develop Nomadic, they divert from what enables their success, which is the robot itself.”

Tanger commends Nomadic’s talent, highlighting that Krishnan is an international chess master ranked as the world’s 1,549th-best player. Krishnan, in turn, boasts that all of the company’s roughly dozen engineers have published scientific papers.

Now, they are diligently working on specific tools, like one that comprehends the physics of lane changes from camera footage, or another that derives more precise placements for a robot’s grippers in a video. The forthcoming challenge, from the perspective of Nomadic and its clients, is to develop similar tools for non-visual data like lidar sensor readings or to unify sensor data across various modes. 

“Handling terabytes of video, processing that against hundreds of 100 billion-plus parameter models, and then extracting their accurate insights is exceedingly challenging,” Bal stated. 

The technology leader in healthcare, CareCloud, has announced that its patients’ electronic health records were compromised by hackers during a recent data breach this month.

The report, submitted to the U.S. Securities and Exchange Commission last Friday, indicated that unauthorized access was detected on March 16 within one of the six environments where the company keeps patients’ medical and healthcare records. According to the company, the hackers had unauthorized access to this medical records storage for over eight hours, although it is still unclear if any data was exfiltrated, or what specific data might have been taken, if at all.

The health technology company stated that it believes the hackers are no longer within its network after the systems were restored on the same day, and has engaged an undisclosed cybersecurity firm to conduct an investigation.

CareCloud did not specify how many individuals were impacted by the breach. The organization offers healthcare technology solutions, including electronic health records storage, for over 45,000 providers, including numerous doctors and healthcare professionals at thousands of hospitals and medical practices, thereby serving millions of patients, as reported in the company’s annual investor report submitted earlier in March.

Providers of electronic health records are lucrative targets for financially motivated cybercriminals, who steal personal information and threaten to release it unless a ransom is paid. In 2024, a ransomware attack by Russian cybercriminals resulted in the theft of a significant portion of America’s health records from Change Healthcare, causing extensive outages and delayed medical care for months.

It remains unclear whether the recent cyberattack on CareCloud led to any data destruction or if the hackers have made any demands to the company. A representative from CareCloud did not return a request for comment. We also inquired about how CareCloud manages patient data, such as whether the company distributes patient data across its six environments or if some of these environments function as backups for others. We will provide updates if we receive a response.

Based on CareCloud’s public internet records, a substantial portion of the company’s files and data are hosted on Amazon Web Services.

CareCloud noted in its SEC disclosure that on March 24, it concluded that the incident was significant enough to potentially impact its business materially and was legally obligated to inform its investors. The company stated that the breach is unlikely to influence its financial situation but acknowledged that its investigation is still ongoing.

Do you have more information regarding CareCloud’s data breach? Do you work at CareCloud and have insight into its security protocols? Reach out to this reporter through an encrypted message at zackwhittaker.1337 on Signal.

The micromobility firm Also, established within Rivian and spun off last year, will collaborate with DoorDash to create autonomous delivery vehicles, the companies declared on Tuesday.

Under the agreement, DoorDash participated in Also’s $200 million Series C funding round, which was spearheaded by existing investor Greenoaks Capital. DoorDash will also secure a position on Also’s board of directors.

This fundraising elevates Also’s total capital to $505 million and values the company at over $1 billion. The startup originated with a $105 million investment from Eclipse, which also supports Mind Robotics — an external industrial AI entity that Rivian established late last year.

Also started as an experimental initiative within Rivian in 2022. The electric vehicle manufacturer initially aimed to produce an electric bicycle and even teamed up with Jony Ive’s design firm LoveFrom, as reported by TechCrunch last year. Also’s initial product is a stylish and unique e-bike, but it has also displayed concepts for compact, pedal-assist delivery vehicles.

Amazon — a significant investor in and collaborator with Rivian — placed an order for thousands of these delivery vehicles late last year. Also has stated these can transport over 400 pounds of packages while remaining compact enough to utilize bike lanes.

The collaboration with DoorDash is the first sign that Also will be creating autonomous iterations of its compact electric vehicles. Rivian founder RJ Scaringe and Also CEO Chris Yu informed TechCrunch last year that nearly any vehicle design was being considered — within reasonable limits.

Both Rivian and Also have indicated that the micromobility venture would utilize the automaker’s technology in its offerings, as well as its retail operations and economies of scale. It remains uncertain if Also will incorporate the autonomy technology that Rivian has been developing.

Rivian is currently set to implement a combination of cameras, ultrasonic sensors, radar, and eventually lidar to enhance its vehicles’ autonomy. Late last year, Rivian announced the development of its proprietary silicon and autonomy computer to make its vehicles more capable of self-driving. Scaringe has already indicated that Mind Robotics would utilize this chip, suggesting that Also may do so as well. TechCrunch has reached out to both Rivian and Also and will update this article if they respond.

There is a possibility that DoorDash will manage the autonomous technology.

The company has its own autonomy division and spent several years developing a robotic vehicle named Dot, which is outfitted with lidar, radar, and camera sensors capable of autonomously navigating roads, bike lanes, and sidewalks. The vividly colored vehicle, resembling cartoonish eyes, can move at speeds of up to 20 miles per hour and is currently delivering in the Metro Phoenix region.

On Tuesday, Roku revealed it is debuting a standalone mobile application for Howdy, its newly introduced ad-free streaming platform that is priced at $2.99 monthly. The app is accessible on both iOS and Android in the United States.

Introduced in August 2025, Howdy boasts a catalog of almost 10,000 hours of material from Roku’s partners, such as Lionsgate, Sony Pictures, Disney Entertainment, Warner Bros. Discovery, and FilmRise, alongside select Roku Original offerings.

The catalog includes titles like “A Haunting in Venice,” “Ice Age,” “Weeds,” and “Kids in the Hall,” in addition to romantic comedies, medical dramas, ’90s comedies, classics, and more.

With this app, Roku claims that subscribers can access Howdy’s content library while on the move.

Costing $2.99 per month, Howdy is presently the least expensive ad-free streaming service available. At its launch, Roku indicated that Howdy was intended to enhance, rather than compete with, premium services.

“In a time when many things are becoming pricier, Howdy is aimed at making premium, ad-free streaming more economical and reachable for all audiences,” stated Gil Fuchsberg, president of Subscriptions, Partnerships and Corporate Development at Roku, in a press release. “The launch of the Howdy mobile app on iOS and Android allows us to expand the service beyond the Roku platform, delivering Howdy’s exclusive value and quality entertainment to an even larger audience.”

The introduction of the app follows a week after Roku announced that Howdy was available on Amazon’s Prime Video. This announcement signified the first expansion of the new streaming service beyond the Roku ecosystem. Last year, Roku and Amazon entered into a significant agreement to share advertising data for their connected TVs.

Roku’s rollout of Howdy occurred two months after the firm invested $185 million to acquire Frndly TV, a streaming service that features live television, on-demand video, and cloud-based DVR.

Howdy also integrates with The Roku Channel, the firm’s free, ad-supported (FAST) streaming service. The Roku Channel is the leading FAST service, surpassing competitors Tubi and Pluto TV. More than 125 million users engage with the platform daily, according to Roku.

Last month, Roku disclosed its 2025 fourth-quarter earnings, reporting a net income of $80.5 million. The company also announced plans to introduce new streaming bundles.

Would you trade your manager for a chatbot? A growing number of Americans are saying yes.

According to a Quinnipiac University poll published Monday, 15% of Americans say they’d be willing to have a job where their direct supervisor was an AI program that assigned tasks and set schedules. Quinnipiac surveyed 1,397 adults in the United States and conducted the poll — which included questions about AI adoption, trust, and job fears — between March 19 and 23, 2026.

Of course, the majority of respondents said they wouldn’t be willing to swap their human boss for an AI people manager. But the use of AI as a supervisor is gaining in popularity, even if one isn’t directly in charge of steering entire teams of people.

Companies like Workday have launched AI agents that can file and approve expense reports on employees’ behalf. Amazon has deployed new AI workflows to replace some of the responsibilities of middle management, laying off thousands of managers in the process. Engineers at Uber even built an AI model of CEO Dara Khosrowshahi to field pitches before meetings with their actual boss.

Across organizations, AI is being used to replace layers of management in what some are calling “The Great Flattening.” Soon, we may start to see entire billion-dollar companies of one, with fully automated employees and executives.

Americans are wary about what that means for their job prospects. The majority of respondents in Quinnipiac’s survey — 70% — said they believe advances in AI will lead to a decrease in the number of job opportunities for people. Among employed Americans, 30% were either very concerned or somewhat concerned that AI would make their job specifically obsolete.