Microsoft Research Just Mapped the Future of Work: What 200,000 AI Conversations Across 923 Jobs Reveal
TL;DR FAQ: What Microsoft Research Found About AI and the Future of Work
➕ Q: What is this new Microsoft Research study about?
A: It analyzed 200,000 real-world conversations between workers and AI tools, capturing 19,265 tasks across 923 distinct job types. This wasn’t theory. It was a direct look at how AI is being used at work right now.
➕ Q: Why does it matter more than previous AI job studies?
A: Most prior studies relied on expert opinions. This one observed real usage. It shows what AI is actually doing, and which job functions it’s affecting most.
➕ Q: What’s the biggest shift from this research?
A: Job titles no longer predict how exposed a role is to AI. The real driver is what specific tasks a worker performs. Two people with the same title can have completely different levels of AI exposure.
➕ Q: Which roles are being reshaped most by AI?
A: Healthcare (non-clinical): Charting, scheduling, and documentation
Technology: Software engineers, analysts, technical writers
Manufacturing: CNC tool programmers (not machinists)
Sales and startups: AI handles research, outreach, and follow-up
Life sciences: Literature reviews and protocol writing
➕ Q: Which roles are seeing the least AI impact?
A: Any work that’s physical, high-risk, or unpredictable
Surgical assistants and patient care roles
Roofers, field service techs, equipment operators
Massage therapists, maintenance crews, cleaners
➕ Q: Is AI replacing jobs?
A: Not directly. AI is replacing or supporting tasks, not full roles. Many jobs are splitting into AI-assisted and human-led components.
➕ Q: What is the SOC code gap and why should I care?
A: Some jobs had dozens of tasks analyzed in the study, others had only a few. That affects how precise the AI exposure ranking is. For better clarity, look at the work being done, not just the job title.
➕ Q: What should employers do now?
A: Redesign processes to mix human and AI capabilities
Break jobs into task-level components
Identify which tasks AI can assist with
Prioritize candidates already using AI tools
➕ Q: What should job seekers do now?
A: Show how you’re already using AI to do your job better
Get fluent in the AI tools used in your field
Emphasize what makes you valuable: judgment, creativity, adaptability
➕ Q: What’s the bottom line?
A: AI is not replacing humans, it’s changing how work gets done. The best hires will be people who know when to use AI and when to rely on their own experience and judgment.
Microsoft Research just published one of the most revealing studies on AI and work to date. Instead of asking experts to speculate about how AI might affect jobs, they analyzed 200,000 real conversations between people and AI assistants. They cataloged 19,265 specific tasks across 923 distinct job types.
This was not theory. It was real-world observation. It revealed how AI is already being used at work – and which roles are being changed the most.
At STEM Search Group, we recruit across engineering, tech, scientific, manufacturing, life sciences, and healthcare fields. This research aligns with what we are seeing every day. Some roles are evolving fast, others are splitting into specialties, and many remain surprisingly unaffected.
Here is what the study tells us – and what it means for hiring and career decisions right now.
Why Job Titles Don’t Tell the Whole Story
One of the clearest findings: job titles are no longer a reliable way to judge how exposed a role is to AI.
Two people with the same title may perform very different tasks. One might be using AI to streamline 80 percent of their work. The other might not use AI at all.
In 40 percent of cases, what the user originally asked AI to do was not what the AI ended up doing. That shift suggests tasks are already being reshaped to match what AI is capable of – even without deliberate planning.
Sector-by-Sector: What We Are Seeing in the Field
We mapped the study’s findings to the industries and job functions we work with every day.
Technology: Speed Is the New Superpower
Roles like software developers, analysts, and technical writers ranked high for AI exposure. But these workers are not being replaced. They are accelerating.
AI supports tasks like code writing, bug detection, documentation, and solution searches. The best engineers we work with treat AI as a second brain, not a crutch.
We now ask every technical candidate, “How are you using AI today in your work?”
Manufacturing: The CNC Divide
CNC Tool Programmers were near the top of the AI exposure list. AI can write toolpaths, optimize G-code, and even simulate machining processes.
But the machinists running those programs? They ranked among the least affected. Their work relies on physical setup, real-time judgment, and human inspection.
Forward-thinking manufacturers are splitting these into distinct roles – AI-assisted programming on one side, human-driven execution on the other.
Sales and Startups: Quiet Leaders
Sales roles ranked number one for AI exposure. Yes, even above software engineering.
Why? AI can handle a lot of the prep work – drafting emails, summarizing accounts, researching contacts, writing call notes. That lets reps focus more on human interaction and closing deals.
Startups are adopting hybrid roles even faster – blending sales, customer success, and technical writing into AI-supported positions from day one.
Life Sciences: Faster, Not Automated
AI helps with literature reviews, drafting protocols, and structuring experiments. But the core work of discovery – forming hypotheses, spotting anomalies, interpreting data – remains very human.
Research teams are already shifting. AI handles early-stage grunt work, while humans handle the insights and decisions.
Healthcare: Still Human, Mostly
Healthcare support roles showed some of the lowest AI exposure in the entire study. That matches what we see on the ground. Hands-on care, empathy, and clinical judgment are not easy to automate.
Where AI is helping is behind the scenes – documentation, scheduling, charting, and billing. It is freeing up more time for patient interaction, but not replacing the work.
Roles with High and Low AI Impact
High AI exposure examples:
- Sales representatives (services, advertising)
- Technical writers
- Web developers
- Data scientists
- Editors and proofreaders
- Public relations specialists
- Customer service representatives
- Market researchers
- Journalists and news analysts
Common theme: communication, repeatable decision-making, and digital execution.
Low AI exposure examples:
- Surgical assistants
- Roofers and field technicians
- Massage therapists
- Machine operators
- Maintenance crews
- Cleaners and housekeeping staff
- Equipment operators in unpredictable environments
Common theme: physicality, safety, and context-driven decisions.
The SOC (Standard Occupational Classification) Code Gap
Some jobs in the study had many tasks associated with them – dozens in some cases. Others had just a few. This means AI exposure scores are more accurate for roles with rich task data.
For example, “technical writer” had dozens of subtasks analyzed. Something like “pile driver operator” may only have had a few. This gap does not mean one job is more or less important – it just affects how precise the AI exposure score can be.
For hiring managers and job seekers alike, it is better to look at the tasks being performed – not just the job title.
What Employers Should Do Now
If you are hiring:
- Break the role down into real tasks, not just a job title
- Figure out which tasks AI can assist with
- Prioritize candidates who are already working with AI tools
- Redesign workflows to mix AI assistance with human decision-making
If you are job searching:
- Learn the leading AI tools in your industry
- Emphasize human strengths: adaptability, judgment, creativity
- Show how you are using AI to get better results today
The Bottom Line
This is not a story about AI replacing humans. It is about humans and AI reshaping work together.
Some jobs are already evolving. Some will remain stable. The winners will be people who know when to let AI help — and when to take the lead.
At STEM Search Group, we are not just matching resumes to job descriptions. We are building hiring strategies for a world where humans and AI work side by side.
Reference: Tomlinson, K., Jaffe, S., Wang, W., Counts, S., & Suri, S. (2025). Working with AI: Measuring the Occupational Implications of Generative AI. arXiv:2507.07935. https://arxiv.org/abs/2507.07935