“It sounds crazy, and people just don’t believe it.”

AI‘s next inflection point comes into focus

Author

Henry Nothhaft, Jr.
May 30, 2025

Your go-to rundown on AI’s impact on the future of work—delivered every Friday. Each edition highlights three to five must-read stories on everything from job disruption and upskilling to cultural shifts and emerging AI tools—all in a crisp, Axios-style format.

Was this newsletter forwarded to you? Sign up to subscribe.

In the wake of the week…

We are not the only ones raising the alarm on AI’s impact at work. From Tim O’Reilly to Anthropic’s CEO, leading voices are warning of deep disruptions to entry-level jobs and talent pipelines. On campus, educators are reverting to analog methods to confront AI-fueled cheating, exposing gaps in how tomorrow’s workforce is being prepared. Inside the enterprise, AI is strengthening the business case for remote knowledge work, giving distributed teams a new edge. And while nearly everyone now encounters AI online, new Pew data shows few engage with it in depth. The implications for leadership, learning, and trust are beginning to take shape.

Let’s dive in. 👇

AI’s Inflection Point: Leaders, Founders, and Futurists Weigh In

Last Sunday, I argued that AI’s impact on the labor market was already underway and that the implications are vastly underestimated. This week’s editorial cycle confirms we are not alone. A growing and diverse chorus of voices is now grappling with the same concerns, from technologists and journalists to policymakers and CEOs.

Tim O’Reilly, the veteran technologist and publisher who helped define the early web, issued a forceful critique of today’s AI narrative. In his post, AI First Puts Humans First, he condemned Silicon Valley’s growing obsession with using AI to eliminate jobs, calling it “anathema,” morally bankrupt, and competitively shortsighted. He accuses a rising class of tech CEOs and investors of fetishizing human displacement as a business model, rather than building systems that amplify what people can do.

O’Reilly calls for an “AI-native” approach: not applying AI as a cheap substitute for labor, but reimagining workflows to do things that were previously impossible. He argues that the companies who pursue augmentation over automation will be the ones that ultimately win.

Meanwhile, Bloomberg Businessweek ran a cover story by acclaimed author and journalist Brad Stone documenting Silicon Valley’s shift away from AI safety toward AI sovereignty. Tech companies are purging ethics teams, softening their calls for regulation, and framing the race against China as justification for speed over safeguards.

And from inside the industry, Anthropic CEO Dario Amodei issued a direct and chilling warning. He forecasts that AI could eliminate 50% of entry-level white-collar jobs within five years. “Most people are unaware that this is about to happen,” Amodei told Axios. “It sounds crazy, and people just don’t believe it.”

What did we advise on Sunday? Listen to what they are telling you.

Taken together, these shifts are not just technical or economic. As Brookings analyst Darrell West notes, they are setting the stage for a public and political backlash. One that could reshape the trajectory of AI entirely.

By the Numbers

  • 50% — Share of entry-level white-collar jobs Anthropic CEO believes could be eliminated by 2030

  • 10–20% — Potential spike in unemployment within five years, per Amodei

  • 95% — Proportion of code at some Y Combinator startups now generated by AI

  • >30% — Share of code at Alphabet now written by AI

  • 50% — Decline in new grad hiring in Big Tech since pre-pandemic levels

Why It Matters

The mainstream conversation is finally aligning with what many workers already sense. AI is just beginning to transform the labor market in ways that will reach far beyond productivity tools or software upgrades. The changes are structural. They come at a time when political leaders are retreating from oversight and stepping away from strategic discourse and active governance.

A new essay by The New Yorker’s Joshua Rothman captures the deeper tension: the AI debate has become a clash of worldviews, not just technologies. Between dire forecasts like AI 2027, (featured in the April 4th installment of this newsletter) and grounded counterpoints like AI as Normal Technology, the result is confusion, not clarity. And when decision-makers encounter disagreement at the top, they often choose delay. That is how we drift into systems we can no longer steer.

Leadership Insight

This is not Don Quixote tilting at windmills. This is the real conversation happening at the highest levels of AI, policy, and business.

Leaders should now ask: Are we using AI to reduce costs or to increase capability? Are we mapping augmentation opportunities or quietly trimming talent? And are we building systems that serve people or systems that people serve?

These are not only ethical questions, but strategic ones, with significant and long-term consequences. Employees are watching closely. In times of uncertainty, they look for clarity, direction, and a point of view. Even without all the answers, transparency matters. Leaders who articulate a position, however provisional, build trust. Those who stay silent create a vacuum.

There is a strong argument that organizations that invest in human-centered AI design, workforce transition plans, and upskilling infrastructure will outperform those who race to zero.

The Bottom Line: The path forward will not be binary. AI will enhance productivity in ways that inevitably reshape staffing models and dampen job growth, even for companies acting in good faith. But intention matters. The organizations that build with a human-centered approach, that use AI to extend capability, will be the ones that endure. Performance, innovation, and trust follow when people remain central to the design.

As AI Redefines Learning, Employers Should Pay Attention

This year’s college seniors arrived on campus before ChatGPT. They are leaving as AI natives. Along the way, many have quietly handed over papers, problem sets, and take-home essays to the bots. The result: a crisis of assessment, a resurgence in analog solutions like blue books, and a deeper reckoning about what school is for.

In recent months, instructors across the country have turned to oral exams, in-class writing, and monitored drafts to regain confidence in student authorship. Blue book sales at major universities are up between 30 and 80 percent. This marks a shift from the early days of AI in education, when the focus was on detection. Use is now widespread. The challenge has moved to managing how, when, and why students engage with these tools.

A survey from Elon University and AAC&U shows that 56% of academic leaders believe their institutions are unprepared to equip students for the AI era. Many acknowledge that students should be learning with AI. Yet policy remains fragmented, and instructors often work without guidance.

That may be changing. A new executive order calls for nationwide AI training for educators and early exposure in classrooms, signaling a federal response to what some now call an “AI literacy arms race” with China. While professors debate the ethics of student use, national leaders are beginning to treat AI fluency as a core competency

And the bigger issue may be one of purpose. As Professor Dustin Hornbeck notes, AI is eroding not just how students demonstrate knowledge, but why they engage at all. In an age in which machines can simulate competence, schools risk hollowing out the deeper human functions of education: curiosity, collaboration, disagreement, and meaning-making.

By the Numbers

  • 90% — College students who used ChatGPT for assignments within two months of its release (New York Magazine)

  • 50–80% — Surge in blue book sales at schools like Texas A&M, Florida, and UC Berkeley (WSJ)

  • 56% — Higher ed leaders who say their institutions are not ready to prepare students for AI (AAC&U, Elon)

  • 1 in 4 — Teens ages 13–17 using ChatGPT for schoolwork (Pew)

  • 66% — Leaders who believe AI will reduce students’ attention spans (AAC&U, Elon)

Why It Matters

Today’s students are tomorrow’s workforce. This is the first generation to be shaped by AI while still in school. Their expectations, instincts, and learning styles will define how the next wave of employees interact with technology. They are already learning to delegate to AI, prompt creatively, and navigate the gray space between enhancement and dishonesty.

This shift will not just affect productivity. It will reshape hiring, onboarding, and skills development. Organizations will need new frameworks to assess readiness, validate expertise, and ensure critical thinking persists in a world of auto-complete cognition.

Leadership Insight

The institutional lag in education today will become a capability gap at work tomorrow. Leaders should anticipate this shift. That means adapting hiring assessments, revisiting entry-level role design, and investing in training programs that reintroduce skills no longer reinforced in school. It also means modeling the kind of integrity, clarity, and human judgment that schools are struggling to preserve.

The Bottom Line: Like it is with work, AI is transforming education from the inside out. Students are learning differently, and that difference will show up in how they work. Employers who understand this shift will be better equipped to support and retain the next generation of talent.

Remote Work Recharged by AI

The death of remote work has been greatly exaggerated. A new wave of research and executive insight suggests that AI may be the thing that gives flexible work its second wind.

Atlassian, Coinbase, and Zillow are doubling down on distributed teams not just for employee morale, but because remote-first environments are becoming ideal testbeds for AI adoption. As Atlassian’s Annie Dean puts it, “AI doesn’t know what’s happening at the water cooler.” It knows what is captured in digital workflows, not hallway conversations.

From automated meeting summaries to self-directed learning, AI tools are streamlining knowledge-sharing and reinforcing performance across distance. Remote environments are, by necessity, already digital-native, making them more compatible with AI augmentation.

By the Numbers

  • 92% of Atlassian employees say the ability to work from anywhere helps them perform at their best

  • 60% of U.S. remote-capable workers prefer hybrid; only 8% prefer fully on-site work (Gallup)

  • 76% of hybrid workers cite better work-life balance as a key advantage

  • 4–5 months — length of social bond durability after Atlassian off-sites

Why It Matters

AI’s deep integration with the digital workflows of modern knowledge work is prompting a reassessment of remote models, just as many companies move to declare work-from-home a relic of the COVID era. Rather than forcing a return to office, leading companies are rethinking their physical footprint and designing hybrid collaboration around trust, tooling, and intentional gathering. As AI handles more coordination and documentation, proximity becomes less essential and flexibility becomes more valuable.

Leadership Insight

The distributed model reflects a structural shift aligned with the digital future of work, offering more than lifestyle flexibility. The same leaders evangelizing a new age of AI-powered productivity, balance, and freedom should explain why so many employers still insist on chaining workers to desks in commercial office space.

The Bottom Line: Companies that integrate hybrid work models with AI deployment strategies are seeing measurable gains in retention, speed, and alignment. The locus of productivity has shifted from physical office space to distributed, digitally enabled environments.

Extra Credit

For the overachievers: These are the stories that didn’t crack the top group this week but are too important to ignore—quick hits on what’s happening and why it matters.

The Trust Problem at the Heart of AI Agents

Key Takeaway: Agentic AI tools promise new levels of efficiency, but their autonomy raises hard questions about whose interests they really serve.

Why It Matters: Without fiduciary safeguards, transparency, or independent oversight, personal AI agents could quietly prioritize advertisers, developers, or sponsors, eroding user trust before it is ever fully earned.

Source
Harvard Business Review: Can AI Agents Be Trusted [metered paywall]

From Laid Off to Leveled Up: How One Exec Used AI to Land His Next Role

Key Takeaway: Mark Quinn lost his job to AI, but turned to the same technology to supercharge his job search, building a custom GPT agent that helped him land a new leadership role.

Why It Matters: Quinn’s experience captures the paradox facing today’s workforce. AI is eliminating roles once seen as secure, but it is also becoming essential to navigating what comes next. Learning to use the tools is a prerequisite for relevance.

Americans Are Surrounded by AI Mentions Online—But Few Dive Deeper

Key Takeaway: New Pew data shows that 93% of U.S. adults encountered AI-related content during web browsing in March 2025, yet only 8% read a news story that discussed AI in meaningful depth.

Why It Matters: The ambient presence of AI in web experiences, from shopping summaries to sidebar mentions, masks a major awareness gap. While exposure is nearly universal, substantive engagement is rare, raising questions about how informed the public truly is about the forces reshaping work, media, and society.

The Consciousness Question Creeps Closer

Key Takeaway: As researchers and technologists debate whether AI might someday become conscious, or already has, new investigations are reframing what machines are and what it means to be human.

Why It Matters: The mere possibility of machine consciousness, even if still remote, has the power to reshape our relationships with technology, our expectations of AI systems, and our moral priorities. The line between simulation and sentience is still science fiction, but increasingly, the question is being asked in serious places: what if?

This edition of The Big Shift: AI @ Work may have been edited with the assistance of ChatGPT, Claude, Copilot, Gemini, Perplexity, or none of the above.

Want to chat about AI, work, and where it’s all headed? Let’s connect. Find me on LinkedIn and drop me a message.