Contrary Research Rundown #138

Unpacking the AI jobs apocalypse, plus new memos on Modal, Synthesia, and more

Contrary is hosting NYC Tech Talk on June 9th, featuring eng leads from Clay, Graphite, Hebbia, and Moment. An evening built by engineers for engineers — each company will live demo their latest product features for leading builders in New York.

Please register to attend here!

Research Rundown

Dario Amodei, the CEO of Anthropic, made a prediction this week that AI would wipe out 50% of entry-level white collar jobs within the next five years, potentially bringing the unemployment rate to 20%. That voice of warning echoed around in many such reports this week from The New York Times to Fortune, Time, the Financial Times, and a host of others. Meanwhile, Mark Zuckerberg has said that AI is soon to replace mid-level software engineers and other companies are developing tools “for a 0-engineer future.”

In one post, Kim Isenberg, the author of Superintelligence, outlined the two different reactions to the evolving details around the coming unemployment wave that could be caused by AI:

“The positions on the current situation regarding AI and the labor market can be divided into two camps: one do not really believe that we are on the cusp of an unprecedented world of work in which robotics and AI work better and more effectively than humans. Their supposed argument: (1) Technology has always created new jobs for humans, (2) Robots have been around for decades and will not be able to replace humans this time either, (3) In reality, the technology of AI and robotics is not that revolutionary, it is just artificially inflated to generate hype and therefore money. These are essentially the arguments put forward as to why everything will probably remain largely as it is, with a few technical and productive improvements.

The second camp, on the other hand, is of the opinion that with AI and robotics we have a technology that is qualitatively new this time. (1) Because AI (in general) reaches human standards intellectually and, as agentic AI, can now and in the future take over many and soon all activities on the computer (“white-collar jobs”) (2) AI is the breakthrough that was necessary to finally raise robots to human quality, in that they can now be trained for all eventualities and carry out delicate activities, so to speak; (3) as a consequence (of 1 and 2), humans as wage laborers are inferior to robotics and AI in the competition for efficiency and are therefore being replaced.”

Some believe that, “even if the AI models were to stop improving, they are already at a level where they are already turning the entire working world upside down.” The discussion from there evolves into what is to be done. As Amodei says, “you can’t stop a train by stepping in front of it; the only thing you can do is steer it.” The proposals span from universal basic income, or an AI tax, maybe with a shared dividend from AI companies that can be distributed to people. Isenberg’s biggest concern is not just the generational labor replacement, but the fact that there is no plan.

Unpacking this potential risk is a difficult needle to thread because it can be immediately pulled into extremes.

You might feel skeptical of 20% unemployment in the next five years given unemployment is currently near all-time lows and even at the peak of COVID, when the world shut down, unemployment topped out at ~14%. But some will feel that means your skeptical of AI as a technology or you’re just not appreciating the pace and magnitude of improvements.

You might feel doubtful of AI’s ability to wipe out an entire swath of roles when you hear about examples like Klarna where the CEO went full-press and replaced 700 customer service agents with AI, only to then reverse course when quality plummeted. But you might be called a Luddite, unable to come to terms with the unstoppable force of AI.

You might hesitate at the focus on job destruction, when very rarely are people able to dream about the potential opportunities that the technology will then unlock. You read things like the World Economic Forum’s Future Jobs Report in January 2025 that estimates, globally, the potential for the creation of 170 million new jobs by 2030, though offset certainly by 92 million existing jobs that will disappear.

You hear the cry of rampant unemployment coming for everyone while also hearing about the mounting expected labor shortages across skilled and non-skilled roles alike, from nursing to hospitality, retail, education, manufacturing, construction, and on and on. 2 million manufacturing jobs here, 1 million construction jobs there; it starts to add up. In five years we’ll go from no robots as nurses to… all nurses as robots?

Palmer Luckey, the founder of Anduril, makes a critical point. Despite the massive gains of AI, despite the imminent potential of AGI or artificial super intelligence (ASI), there is still a massive limiting factor: human inertia.

“The limiting factor for the adoption of ASI is not going to be the technology itself. It's going to be human inertia. How long does it really take to penetrate the world and actually take over these things? Look how long it's taken Square to replace old point of sale contracts where Square was strictly the better product. Lower fees, better compatibility, better terminals, lower cost. And it still takes years for them to try and penetrate and actually get spread across the network. I feel like it's unlikely that there will be [one] winner and they'll take over every industry so fast that nobody has time to react.”

Does this mean that AI is all hot air and hype? Absolutely not. Anduril is the first to say that AI is integral to everything they do. But at the same time that Silicon Valley writ large is spending so much time focusing on the exponential capabilities of AI, they’re also pitching SaaS companies into industries that are often still run on “pen and paper.” Do you think the reason that pen and paper has persisted across literally billions if not trillions of dollars of economic value is because the operators in those categories were holding off on the software in favor of the AI-enabled robotics?

The reality is that progress will, in fact, be exceptional. But the complexities of human inertia, social systems, organizational infrastructure, and consumer expectations have consistently been a barrier for technological progress. That doesn’t mean that progress doesn’t happen. But it means that society typically pushes back, forming technology around itself in a way that can certainly cause some short-term pain, but should ultimately instill long-term optimism.

Synthesia is a provider of AI-generated video for enterprise use, offering a browser-based platform optimized for training, onboarding, and internal communications. To learn more, read our full memo here and check out some open roles below:

• Senior Full Stack Engineer - Remote (EU, UK or Switzerland)

• Engineering Manager (AI dubbing and Localisation) - Remote (EU, UK or Switzerland)

Modal offers a serverless compute platform aimed at addressing common pain points in AI infrastructure, including slow deployment times, high operational complexity, and inefficient cost structures. To learn more, read our full memo here and check out some open roles below:

• Product Engineering Manager - New York, NY

• Cloud Quant - New York, NY

Check out some standout roles from this week.

• Hinge Health | San Francisco, CA (hybrid) - Software Engineer II (Fullstack), Site Reliability Engineer II, Senior Data Engineer (Claims Data), Senior Product Manager (MarTech)

• Discord | San Francisco, CA - Senior Data Engineer, Product Manager (Core Product), Senior Full-Stack Engineer (Ads), Software Engineer (Growth)

• Ripple | San Francisco, CA or London, England - Senior Software Engineer, Manager (Software Engineering), Senior Security Engineer (Detection and Response)

• Reed Hastings joins Anthropic’s board to guide responsible AI growth. The Netflix co-founder was appointed by Anthropic’s Long-Term Benefit Trust to help oversee its $2 billion business while maintaining safety and social responsibility.

• AI is forcing colleges to redefine academic integrity. With ChatGPT-like tools now common in student work, and even in grading, universities face a crisis of credibility. Bloomberg argues colleges must tighten policies, increase in-class and oral assessments, and set clear AI-use norms to preserve learning and civic education.

• Walter Isaacson predicts xAI-Tesla merger driven by real-world AI vision. The biographer of Elon Musk says he is likely to combine xAI with Tesla, citing Musk’s belief that true AI requires training on both language and sensor data from Teslas and Optimus robots, enabling a step-change beyond current chatbot models.

• Carl Pei, the founder of Nothing, predicts a one-app phone future with AI pushing smartphones toward a single intelligent OS that personalizes, automates, and proactively suggests actions for users. Pei argues that apps will gradually fade as phones evolve into context-aware agents, though adoption will take 7–10 years.

• Palantir-backed “Warp Speed for Warships” targets US naval production crisis. In a call to revitalize American shipbuilding, Mike Gallagher outlines how AI-enabled tools like Palantir’s Warp Speed platform are accelerating throughput at firms like Anduril and Saronic. With China outbuilding the U.S. Navy, the initiative aims to unify digital procurement, predictive maintenance, and workforce training into a national industrial surge.

• North Korean IT workers infiltrate U.S. companies using “laptop farms.” A former waitress in Minnesota helped North Koreans pose as remote tech workers using stolen U.S. identities and her home as an operational hub. The scheme funneled over $17 million to North Korea and compromised firms across defense, media, and tech.

• Anduril and Meta team up on military XR headsets in Luckey’s full-circle moment. Under the U.S. Army’s $22 billion program, Meta’s Reality Labs will support hardware and Llama AI models while Anduril takes on the prime role in the contract, and integrates battlefield software. The project revives Palmer Luckey’s original AR headset vision called EagleEye years after his ousting from Facebook.

• Retool unveils autonomous AI “Agents” that automate enterprise workflows. The new product connects directly to APIs, databases, and internal tools to execute tasks without human input—automating 100 million hours of work and saving $6 billion to date. Retool says Agents are priced at $3/hour and aim to replace 10% of U.S. labor by 2030.

• Scientists have 3D printed tissues inside the body using ultrasound. A Caltech team developed a technique to bioprint structures directly in living tissue using a novel ultrasound-activated bioink. The system enables printing drug depots, biosensors, and custom implants without surgery, marking a major step toward non-invasive regenerative medicine.

• At BUILD, Microsoft showcased Copilot as a “peer programmer,” but internal tests on the complex .NET codebase revealed major limitations. Transparency was refreshing—but also highlighted how immature agentic tools remain in actual engineering environments. Gergely Orosz wrote more about it here.

• Dan Shipper also sat down with GitHub CEO, Thomas Dohmke, to unpack the details. While GitHub Copilot has 15 million users, it faces pressure from faster-moving rivals. Despite its scale, Copilot has struggled to outpace startups like Cursor and Windsurf. Dohmke said their new autonomous coding agent—embedded directly into GitHub—aims to close that gap by serving both AI-native and enterprise devs.

• The Browser Company may sell or open source Arc amid pivot to AI browser Dia. CEO Josh Miller said Arc’s complexity limited its appeal and development has ceased. But open sourcing is unlikely soon, as Arc’s internal SDK also powers Dia, the company’s new AI-native browser now in alpha.

At Contrary Research, we’ve built the best starting place to understand private tech companies. We can't do it alone, nor would we want to. We focus on bringing together a variety of different perspectives.

That's why applications are open for our Research Fellowship. In the past, we've worked with software engineers, product managers, investors, and more. If you're interested in researching and writing about tech companies, apply here!