There is a lot buzz in the news about AI replacing jobs. As someone who keeps up with technology, I recognize that today’s tools are powerful enough to transform much of the work we do. But when I talk to people who work at knowledge based jobs it seems that most organizations still aren’t using AI in a deep, operational way. So I did a little research to answer a simple question: what’s actually happening right now? How is AI being used in the workplace?
Two things stuck out:
1) “Basic AI” is normal
Basic AI is everywhere. This is the day-to-day stuff. Most people are using AI to do things such as:
-
writing and polishing emails
-
cleaning up grammar and tone
-
summarizing notes and long documents
-
generating first drafts and quick outlines
Hardvard Business Review found that 88% of survey respondents report regular AI use in at least one business function. According to Gallup regular AI use includes things such as consolidating information/data (Summarize a long doc), 41% generating ideas (Brainstorm), 36% learning new things, 34% automating basic tasks (Draft/rewrite emails), and 20% identifying problems.
Most of this is personal productivity, not “the company changed how work works.” It’s me using a tool, not the business running differently.
2) “Advanced AI” is moving slower
Now zoom out from “help me write faster” to “help us run the business differently.” This is where AI starts touching real workflows such as routing and triaging tickets, automating multi-step processes, searching internal knowledge reliably, drafting + reviewing + handing off work across systems.
The recent NBER paper is blunt about what firms say about this type of AI use: more than 90% of firms report no impact on employment, and 89% report no impact on productivity
This isn’t because leaders are clueless. It’s because the “advanced” layer is where the hard problems live. Barries include the following:
-
Trust (reliability) : AI can be confidently wrong, and that’s expensive
-
Data/privacy: companies can’t just dump internal info into anything
-
Change management: people + process beats tech demos every time. People are generally resistant to change. Even the best AI tool won’t help much unless people are trained and the workflow is redesigned so everyone knows how and when to use it.
-
Integration: legacy systems are a mess
-
Compliance: regulated industries move carefully for a reason
KPMG’s trust research is a gut check: only 46% say they’re willing to trust AI, and a lot of people use AI outputs without evaluating accuracy. That’s exactly why companies hesitate to automate important workflows.
My thoughts on the near future
I believe the next 12 months will see the following:
-
AI becomes default inside docs/email/meetings (it already is for the most part)
-
more policies and guardrails as leadership catches up and new roles arise around AI ops, governance, quality, risk arise
-
AI begins to augment jobs as a coworker.
Two to three years out:
-
more workflow automation
-
business will see a meainingful boost in productivity from AI
- potential for knowledge base job displacement.
My take away is that advanced AI use is coming, but it has to earn trust first. Productivity boost and job displacement may become real within the near future, but hopefully new roles emerge.