An overhead composition of blank paper sheets fanned out in a semicircle on a dark concrete surface, soft golden lighting

🧭 Career Digest Tuesday, 16 June 2026

Career & Future — June 16, 2026

Workers are spending over 6 hours a week babysitting AI tools, Forrester finds employers regretting early AI layoffs, and the 'AI won't replace engineers' essay goes viral. Plus Microsoft AI chief walks back his white-collar apocalypse prediction.

🔍 DIGEST SUMMARY

Four days of workforce and career stories you missed. The headline is the new hidden labour: Business Insider reported that workers are now spending over 6 hours a week “botsitting” AI tools — prompting, fixing, and correcting AI output that is not yet autonomous enough to run unsupervised. The piece is the first mainstream reporting on what the consultancy world has been calling “the hidden labour of AI deployment.” Microsoft AI chief Mustafa Suleyman walked back his most famous white-collar-apocalypse prediction, Forrester published data on employers regretting early AI layoffs, and a viral essay argued “why AI hasn’t replaced software engineers, and won’t.” Homebrew 6.0 dropped. The through-line: the AI labour story is not “AI replaces humans” — it is “AI replaces some tasks, creates new tasks humans must do, and the new tasks are not yet valued or paid for.”

6+ hours/week “botsitting” — the new hidden labour of AI deployment.
Suleyman walks back white-collar apocalypse — Microsoft AI chief’s pivot.
Forrester: employers regret AI layoffs — the first survey data on post-layoff sentiment.
“AI hasn’t replaced software engineers” — the canonical essay on why the engineering labour market is structurally different.
Homebrew 6.0 — the canonical Mac developer tool drops, signalling the post-ChatGPT developer ecosystem.
Microsoft AI layoffs in NZ context — the local angle on the global trend.
The skill premium shift — what’s now paid for in AI-augmented work.

Workers Are Spending 6+ Hours a Week “Botsitting” AI

The biggest career story of the week, by reach: Business Insider’s piece on the “botsitting” hidden labour reports that knowledge workers are now spending an average of 6.2 hours per week on AI-related tasks that are not yet autonomous — prompting, fixing, correcting, and validating AI output. The story earned 279 HN points (221 comments). The figure breaks down roughly: 2.1 hours on prompting and re-prompting, 1.8 hours on validating AI output, 1.3 hours on correcting AI mistakes, and 1.0 hour on the meta-task of managing the AI tool itself (context setup, plugin management, billing reconciliation). The 6.2-hour figure is the first hard data point on the labour cost of partial AI autonomy, and the cultural implication is significant: AI is not yet a productivity multiplier at the worker level. It is a productivity tool that demands a new kind of labour, and that labour is not yet priced in.

Mustafa Suleyman Walks Back the Apocalypse

Microsoft AI chief Mustafa Suleyman — the most public voice on the white-collar-AI-apocalypse thesis — has publicly walked back his most famous prediction. The shift, reported by The Verge and covered across multiple outlets, is the first time a senior figure at a frontier lab has publicly softened the AGI-is-imminent framing. Suleyman’s new position: AI will augment more jobs than it will replace, and the labour displacement will be slower than the 18-month window he gave in 2024. The Forrester data below is the empirical support for the new framing. For workers, the practical implication is that the “learn to code / retrain for AI / pivot to AI safety” advice that was everywhere 12 months ago is now being walked back by the people who were pushing it hardest.

Forrester: Employers Regret Early AI Layoffs

Forrester published survey data showing that employers are regretting the AI-driven layoffs of 2024-2025 — particularly the round of “AI replaces X% of headcount” announcements that came in Q4 2024. The data, summarised in a free report and covered by Neowin and India Today, shows that 67% of HR leaders now say the AI-driven reductions of 2024-2025 either did not produce the expected cost savings or actively reduced productivity because the work had to be re-hired or contracted out. The Forrester data is the first empirical evidence that the AI-labour-replacement thesis at the firm level has not held up in practice.

”Why AI Hasn’t Replaced Software Engineers, and Won’t”

The most-shared engineering-career essay of the week, by an anonymous author at Normal Tech, argues that the structural reasons AI has not replaced software engineers are durable and will not change. The argument in summary: software engineering is not a single task but a portfolio of tasks (requirements gathering, system design, code review, deployment, monitoring, on-call, stakeholder management) and AI excels at the central task (code generation) but is net-negative on the rest. The essay earned 311 HN points (356 comments). For NZ software engineers, the practical implication is that the current demand softening is real but not catastrophic, and the structural premium on full-stack engineering — not just code generation — is the durable thing.

Homebrew 6.0

Homebrew — the canonical Mac developer package manager — shipped 6.0 on June 11. 1,465 HN points, 360 comments. The release notes mention AI-assisted dependency-resolution as a new feature, and the project’s “AI policy” page is one of the clearest in the open-source world: Homebrew accepts AI-assisted contributions but requires the contributor to disclose AI use and to be the human who understands and stands behind the contribution. The Homebrew 6.0 release is the most visible signal yet that the major OSS projects are converging on a “AI-assisted is fine, AI-replacement is not” norm.

Microsoft AI Layoffs in the NZ Context

The local angle: the Microsoft 2024-2025 AI-driven layoffs hit roughly 3% of the company’s NZ workforce, per industry sources, and the regional job market for senior engineering roles has softened but not collapsed. The local consulting sector is now hiring AI integration engineers at a 15-20% premium over the equivalent 2023 roles, but the volume of new roles is roughly 60% of the volume lost. The net effect for an NZ engineer in 2026 is a labour market that is tighter than the global headlines suggest but still structurally weaker than 2022-2023. The advice from local recruiters is consistent: pivot to AI integration work, do not wait for a “generalist engineering” recovery.

The Skill Premium Shift

The structural shift in 2026 is the rise of “AI integration” as a paid skill premium. Workers who can ship AI-augmented workflows — not just use AI tools, but design the workflows around them — command a 15-25% wage premium in the consulting, finance, and software sectors. The premium is highest in roles where the AI tool does not yet replace the human but changes what the human is paid to do. The skill premium is real, durable, and is the most important career signal in the 2026 labour market.

Three through-lines. First, the AI-labour story is no longer a story about replacement. The “botsitting” data, the Suleyman walk-back, the Forrester regret data, the engineering-career essay — all of them are converging on the same point. AI is a productivity tool that demands new labour, not a labour-replacement technology. Second, the workers who are winning in 2026 are the ones who can design AI-augmented workflows, not the ones who can use AI tools. The Homebrew policy is the OSS community’s version of the same signal. Third, the labour-market recovery in 2026 is uneven. Consulting, finance, and AI integration roles are paying more than 2023. Generalist engineering and entry-level white-collar roles are still soft. The advice is to specialise toward AI integration rather than wait for a generalist recovery.