In partnership with
Welcome back. This edition unpacks a new Dropbox survey on the busywork AI is creating. (If you were hoping for me to weigh in Meta’s Muse, hit reply.)
And for paid members: Amit Sevak, CEO of testing giant ETS, tells Macro Talk about an upcoming assessment product from newly acquired ACT.
Upgrade your free subscription to unlock the full Macro Talk experience.
Forwarded this email? Sign up here.
The Rising Tide:
AI Busywork
Screenshot: Noam Blum/X
One of my favorite future-of-work stories from my time at Axios was about how tech can make work harder.
Gartner told me at the time that workers were using an average of six to eight apps to perform a single business process.
A separate Harvard Business Review story suggested that workers were switching from app to app, and website to website, some 1,200 times a day — a “toggling tax” that added up to about 9% of their time at a job.
Well this week, a Dropbox-sponsored study suggests AI is growing this category of busywork.
“Nearly three-quarters (74%) manually move AI-generated work between applications at least three times on an average AI workday. For 30%, that happens six or more times,” the report reads.
And perhaps more relatable: 65% use three or more apps to finish a typical AI-assisted task — while a mere 11% of respondents say AI produces work that could be considered fully finished.
Methodology: The 504 U.S. participants of the survey use AI and work in B2B roles ranging from IT, engineering, sales, marketing, finance, ops, HR, and customer support.
What this means for workers
You’re not imagining things. Working faster in one area doesn’t mean working less.
And if only 11% of people say AI output is truly finished, the humans doing the finishing aren't going anywhere yet. Show your boss the stat.
What this means for executives
“AI can generate an answer in seconds, but getting that output across the ‘last mile’ into finished work still requires the right context, verification, people, permissions, and workflows. Today, that often means moving content between apps, sharing it with team members, and getting it ready for review,” Dropbox says.
Every handoff between apps is another place where context, version history, and accountability can slip — so budget for the oversight.
The upside: companies already built for oversight may have better checks already in place. As former House Financial Services Committee Chairman Patrick McHenry put it at FIS's Emerald conference earlier this year: "Regulated institutions that know their customers, understand communities, and can be examined, audited, and held accountable — that is not a burden. It's a moat."
A message from our sponsor:
Some teams never seem to stop moving. They're on Attio, the agentic CRM.
Every customer signal is captured in one shared context layer, always current and compounding. Agents and workflows build pipeline, chase every buying signal, and move deals forward, an always-on revenue engine running alongside your team.
With Attio, you’ll get:
Leads automatically prioritised and routed to the right rep
Expansion and risk signals caught the moment they land
Follow-ups written in your voice, already there when you arrive
Teams like Parallel, Turbopuffer, and Wordsmith build on Attio. Are you one of them?
Strategy Talk:
ETS Has Big Plans for ACT
"AI in its current form" should not be in elementary school classes, ETS CEO Amit Sevak tells me in a new episode of the Macro Talk podcast.
But “it becomes increasingly critical” to define the appropriate uses for older students — particularly in high school — because they’ll be better prepared for jobs, he argues.
“There are risks, but the benefits, I think, outweigh the risks in terms of readiness of kids for the future,” says Sevak.
Why his perspective matters
ETS — known for its TOEFL English Language Test and the GRE — is the largest nonprofit education testing organization in the world, it says.
As CEO, Sevak is charged with adapting the $800 million business to an increasingly fragmented world of learning and work, amid a continuing decline in standardized testing participation.
How he interprets change helps shape how ETS responds — from the products it builds to the policies it influences — with consequences for the educational and professional lives of tens of millions of people every year.
This summer, two years after losing its contract to administer the SAT, ETS acquired standardized test provider ACT, Inside Higher Ed reported.
“We bought the ACT not just to have another test in the ETS portfolio,” says Sevak. “We bought it because we think there's an amazing opportunity right now to significantly modernize the whole portfolio of ACT assessments.”
ACT runs WorkKeys for example. It’s a skills-focused testing platform used by thousands of U.S. employers, per Inside Higher Ed.
The fork in the road
ETS’s pivot to skills and workforce readiness is supported by demand from employers who want to find the right workers based on evidence of what they can actually do.
From the National Association of Colleges and Employers (NACE):
“The use of skills-based hiring by employers is growing, according to results of NACE’s Job Outlook 2026 survey. Among survey participants, 70% report using skill-based hiring, up from 65% last year. Furthermore, among this year’s employers, 71% use this approach at least half of the time.
The stages in the recruiting process when employers use skills-based hiring most often are during interviewing (87%) and screening (65%). The latter is especially noteworthy as this approach has partly supplanted GPA as a screening tool. In 2019, nearly three-quarters of employers — 73% — screened candidates by GPA; this year, just 42% are doing so.
Still, skill development among candidates is important whether or not employers use GPA: 73% of employers that do not screen by GPA and 67% of employers that do screen by GPA consider if a candidate has demonstrated proficiency in key competencies when making hiring decisions.”
Inside school walls, Sevak also sees a role for smaller, more frequent "mini assessments," also called formative or interim assessments, that work as "a more adaptive quiz mechanism."
“The half-life of a skill is shifting, and therefore we've got to have validation of skills at different moments in time for different roles,” he says.
But there’s another area ETS is focused on growing
Sevak tells me he already has plans to offer a new product from ACT centered on
Subscribe to Macro Talk to read the rest.
Become a paying subscriber of Macro Talk to get access to this post and other subscriber-only content.
Upgrade