Nearly each UC vendor pitch in 2025 and 2026 has led with the identical promise: AI will save time, automate duties, and free workers to deal with higher-value work. Assembly summaries, message drafts, automated workflows: the productiveness positive aspects sound inevitable.
However a brand new eight-month examine from UC Berkeley researchers, printed in Harvard Enterprise Overview, has discovered the other. Monitoring roughly 200 workers at a US tech agency, professors Aruna Ranganathan and Xingqi Maggie Ye found an AI productiveness paradox: discovering that AI instruments didn’t cut back work however persistently intensified it:
“On their very own initiative employees did extra as a result of AI made ‘doing extra’ really feel attainable, accessible, and in lots of instances intrinsically rewarding.”
Workers voluntarily labored at a quicker tempo, took on a broader scope of duties, and prolonged work into extra hours of the day. One engineer captured the fact: “You had thought that perhaps, oh, since you might be extra productive with AI, then you definately save a while, you’ll be able to work much less. However then actually, you don’t work much less. You simply work the identical quantity or much more.”
The query for UC leaders is straightforward, then. Is that this already taking part in out inside your collaboration stack?
The Knowledge Says It’s Already Occurring
Impartial analysis from Microsoft, PwC, and Gartner confirms the identical sample at scale — and particularly contained in the UC platforms enterprises are deploying right now.
Microsoft’s Work Pattern Index, launched in June 2025 and primarily based on trillions of Microsoft 365 productiveness indicators, revealed what it calls the “infinite workday.”
It discovered the typical worker now receives 117 emails and 153 Groups messages each day. Staff are interrupted each two minutes: equal to 275 instances per day.
Key findings on AI work intensification:
40% test e-mail earlier than 6am
29% are again of their inbox by 10pm
Night conferences after 8pm, up 16% year-over-year
48% of workers say their work feels chaotic and fragmented
PwC’s 2025 International Workforce Hopes & Fears Survey, in the meantime, which surveyed practically 50,000 employees throughout 48 nations, discovered that 35% of the worldwide workforce feels overwhelmed a minimum of as soon as per week — rising to 42% amongst Gen Z. But solely 14% of employees are utilizing generative AI each day. Satirically, these each day customers report important positive aspects in productiveness (92%), job safety (58%), and wage (52%).
In keeping with Pete Brown, International Workforce Chief at PwC:
“Workers utilizing AI daily are reaping the rewards – larger productiveness, higher job safety and higher pay. However to scale these advantages, companies should transcend coaching. Work itself must be redesigned.”
Gartner’s 2026 Way forward for Work Traits for CHROs lists “AI’s greatest hidden price: your workers’ psychological health” as a top-nine pattern. The analysis agency additionally flags “AI workslop” — low-quality AI output that workers spend hours reviewing and fixing — as organizations’ prime productiveness drain.
Three Methods It Exhibits Up in UC
The Berkeley examine recognized three types of work intensification. Every maps instantly onto dynamics inside unified communications platforms.
Position Creep by way of AI Copilot
As a result of AI copilot instruments fill gaps in data, employees more and more step into duties that beforehand belonged to others. The Berkeley researchers discovered product managers writing code and researchers taking up engineering duties.
When collaboration platforms make it straightforward for anybody to draft workflows, write scripts, or automate processes, position boundaries dissolve. The unintended consequence? Specialists — significantly engineers — get pulled into reviewing, correcting, and guiding AI-generated or AI-assisted work produced by colleagues.
The At all times-On Collaboration Layer
Staff within the examine despatched “one final immediate” earlier than leaving their desk, used AI throughout lunch breaks, and prompted instruments throughout conferences. The conversational interface made beginning a activity really feel nearer to chatting than enterprise formal work.
In Microsoft Groups, Slack, or Webex, the road between messaging a colleague and prompting an AI assistant has collapsed fully. That is exactly why 29% of employees are again of their inbox by 10pm: the friction between work and non-work has been eliminated, and the collaboration platform is the place it occurs.
Agent Sprawl in UC Platforms
As unified communications platforms push agentic AI instruments — Copilot Studio, Zoom AI Companion, Salesforce Agentforce — employees handle a number of autonomous threads concurrently. The Berkeley examine discovered this created “a continuing switching of consideration, frequent checking of AI outputs, and a rising variety of open duties.”
Whereas employees described having an AI “accomplice” that helped them transfer by means of their workload, the fact was fixed context-switching and cognitive load. The extra brokers you deploy, the extra human oversight they require — a brand new administration layer no one budgeted for.
What UC Leaders Ought to Do Concerning the AI Productiveness Paradox
The Berkeley researchers suggest that organizations develop an “AI observe” — intentional norms and requirements round AI use. For UC leaders addressing the AI productiveness paradox, that interprets into 4 concrete actions:
Implement Focus Time on the Platform Degree
Do-not-disturb scheduling, notification batching, and guarded focus home windows exist already in Groups, Zoom, and Slack. However these options solely work in the event that they’re coverage. Configure them organization-wide and make disconnection a cultural norm
Govern AI Brokers Earlier than They Sprawl
As AI copilot and agent capabilities increase throughout your UC stack, set up governance now: who can deploy brokers, what they will act on, and the way outputs get reviewed.
Audit AI-to-Human vs. Human-to-Human Time
The Berkeley researchers emphasize “human grounding” — brief alternatives to attach with others that interrupt steady solo engagement with AI instruments and restore perspective. Shield 1:1s, workforce retrospectives, and unstructured collaboration time.
Measure Outcomes, Not Simply Pace
“Hours saved” is the seller metric. Monitor determination high quality, error charges, rework cycles, and worker wellbeing alongside throughput. The Berkeley examine warns that what seems to be like larger productiveness within the brief run can masks silent workload creep and rising cognitive pressure.
