Ryan Roslansky: Turning AI anxiety into skills for the future of work

Roslansky & Smith (Tools and Weapons) — turning AI anxiety into skills

▶ Watch on YouTube · Microsoft · 41:15

In this episode of Tools and Weapons, Microsoft’s Vice Chair and President Brad Smith sits down with Ryan Roslansky, CEO of LinkedIn and Executive Vice President at Microsoft, for a practical conversation about AI, the future of work, and what people can do now to stay ahead. Drawing on ideas from his new book, “Open to Work: How to Get Ahead in the Age of AI,” Ryan shares a framework for thinking about careers in the era of AI that starts not with job titles, but with tasks.

TL;DR

A 41-minute episode of Tools and Weapons, Brad Smith’s podcast on the Microsoft channel, published 31 March 2026. Smith, Microsoft’s vice chair and president, interviews Ryan Roslansky, CEO of LinkedIn, about Open to Work, the book Roslansky wrote with Aneesh Raman. Five claims:

1. Jobs are tasks, not titles, in three buckets: tasks AI will automate (“Summarize this document”), tasks it augments (“a superpower”), and tasks “so innately human or messy” that AI will not do them, such as getting a disagreeing room to agree.
2. The human skills that were called soft are the hard currency now: “curiosity, courage, communication, compassion.”
3. Skills change fast regardless of AI. On LinkedIn’s data, the skills required for the average job changed by 25% over the past eight years, and LinkedIn expects 70% change by 2030. “Even if you’re not changing your job, your job is changing on you.”
4. One builder instead of four roles. LinkedIn created a builder role that combines product manager, product marketer, designer and front-end engineer, plus an associate builder entry route for people “coming out of college, maybe not even going to college at all.” Product managers now write evals rather than specs.
5. A schism in the industry. Smith names “the emerging great debate”: AI built to “outperform humans at… most economically valuable work” (AGI) against AI built to “empower people to do more.”

Key claims

A framework, not a forecast

Roslansky frames the book as “not as a crystal ball… but more… as a framework” for turning uncertainty into opportunity, and “the future isn’t written.” Smith supports the modesty with an anecdote: he asked Copilot’s Researcher agent to grade well-known technologists’ past predictions, and “almost everybody got between 20 and 30%.”

Three task buckets, and using the middle one

Classify your own day by bucket and ask what share of your job is in each. Roslansky’s bucket-two examples from his own work: Copilot in Outlook summarising a long email debate (“I can save a good ten to 15 minutes on a topic”), and asking it to challenge an important email before he sends it, “as a thought partner.” Moving tasks from bucket one to bucket two frees time for bucket three, which Smith calls “where we’re often the most time constrained,” such as a manager asking colleagues how they are doing.

Careers as climbing walls

People ask LinkedIn for career paths, but “there is no such thing as a linear career path.” Think less about the next title and more about “what adjacent skills or tasks can I be picking up.” He sees individual contributors who once assumed they had to become managers staying ICs, because with AI they can do more. The book’s three questions: why do you work, what do you uniquely do, and where do you want to go.

How jobs are changing

Software engineering is changing, not going away: LinkedIn sees “an increase in the number of software engineer postings.” A product manager still talks to customers, but the next step is now “writing a bunch of evals for the model”: “If this scenario, here’s a good or bad response.” At LinkedIn, siloed roles in software production (PM, product marketer, designer, front-end engineer) required a communication path between them; one builder with AI tools can take a product from start to finish, so the company can test more. “It doesn’t mean anyone’s role is going away. We’re making more builders.”

Outperform or empower

Smith: some in the industry pursue AGI defined as a system that outperforms humans at most economically valuable work; Microsoft’s mission is “technology that empowers people to do more.” Roslansky’s answer stays with the second: AI as “an enabler for us to do bigger and better things as a society,” which needs governments, schools and companies as well as technology firms.

Dynamic-capabilities reading

• digital-sensing/digital-mindset-crafting — the book’s purpose: a framework for individuals to read their own work as tasks and act on it, with curiosity and practice as the habits.
• digital-transforming/redesigning-internal-structures — LinkedIn’s builder role merging four siloed roles, and the associate builder entry route.

How it connects

• Raman & Wood (WorkLab) — the book’s co-author on the same ideas, three months later; together the two pages give both authors’ accounts.
• George (Microsoft) — Microsoft engineers told they are now “product builders”; the same move at the parent company.
• durable-skills — curiosity, courage, communication, compassion.
• expert-generalist — the builder role as the organisational form of the generalist.
• ai-employment-effects — skills churn of 25% and 70%, rising engineering postings, and a non-degree entry route.
• automation-vs-augmentation — the three buckets, and the outperform-or-empower debate.

What was actually ingested

The full episode (41:15, length_seconds: 2475) from the human-curated caption track. The book itself is not in the wiki. No stills: a podcast.

Linked entities and concepts

• Microsoft — publisher (Tools and Weapons), host’s employer, and LinkedIn’s parent. Updated in this ingest.
• durable-skills, expert-generalist, ai-employment-effects, automation-vs-augmentation — updated in this ingest.
• dynamic-capabilities — the two cells tagged above.

Dangling (single-source mention, deferred): Ryan Roslansky, Brad Smith, Aneesh Raman (now named on two sources, never author:), LinkedIn.

Scope and reliability

Two Microsoft executives discussing one of them’s book on Microsoft’s channel; the host is the guest’s colleague and praises the book throughout. The LinkedIn figures (25%, 70% by 2030, rising engineering postings) come from LinkedIn’s own data without method in the episode. The episode is from March 2026, seven months before ingest. Its value is the task-bucket framework and a concrete organisational change at LinkedIn (the builder role and its entry route).