What AI and jobs data does and does not tell virtual assistants
New data on AI and jobs does not say that virtual assistant work is ending. It also does not say that your work is safe. It says that hiring for junior, task-based roles is changing in some US occupations, and that the effect depends on whether AI replaces a task or supports a person. You can use that idea to plan your own skills.
Three data sources, three different questions
1. Stanford: early-career hiring in the US
Researchers at the Stanford Digital Economy Lab (Brynjolfsson, Chandar, and Chen) studied payroll data from ADP. The paper was revised on August 12, 2026, and covers US workers through June 2026. They found no sign of job loss across the whole economy. They did find that employment of workers aged 22 to 25 in AI-exposed jobs is 19% below where it would be if it had kept pace with less-exposed jobs. Experienced workers show no gap of this kind.
The authors say the drop comes mostly from less hiring of young workers, not from more firings. They also found that employment falls where AI mainly substitutes for human tasks. Where AI mainly complements workers, employment is flat or rising.
Limits: The authors call the results early and descriptive. They do not prove that AI caused the change. The data is from US firms that use ADP. It is not about the Philippines, and it is not about freelancers.
2. Philippine industry figures
The Customer Experience Association of the Philippines (CXAP) reported that the contact center and business process management workforce grew about 4% in 2025. It went from 1.62 million to 1.68 million jobs. CXAP projects 1.73 million in 2026. The Freeman reported these figures on June 13, 2026.
Limits: An industry group produced these numbers, and it has an interest in a positive story. The figures describe the large BPO sector. They do not describe independent virtual assistants. Projections are forecasts, not results.
3. Upwork: what clients bought on one marketplace
Upwork’s In-Demand Skills 2026 report (February 4, 2026) used 2025 freelancer earnings on its platform, with demand from the US. Demand for skills that name AI grew 109% year over year. In the customer service and admin support category, general virtual assistance was the most in-demand skill. Upwork reports “consistently strong” demand for it but gives no growth figure.
Limits: This is one company’s marketplace data. Upwork sells the service it measures. It does not cover the Philippines specifically.
What the three sources do and do not tell you
| Question | Answered? | Note |
|---|---|---|
| Is my task exposed to AI? | Partly | Exposure means AI could do part of a task. It does not mean the job is gone. |
| Have jobs like mine disappeared? | No | None of the three sources measures remote VAs in the Philippines. |
| Will my work disappear? | No | Nobody can say. The sources show forecasts and past trends. |
| What kind of work holds up better? | A pattern | Stanford’s data suggests that complement roles hold up better than substitute roles. This is a pattern in one dataset. |
A task-map exercise
- List 10 tasks you did for clients in the last month.
- Mark each task S (AI could mostly do it alone), C (AI helps, but you decide or check), or H (needs human trust, context, or accountability).
- Count your S tasks. These are the tasks most likely to be priced down or done by the client with AI.
- Pick one S task. Write how you could move it toward C. Example: instead of only entering data, you check the data, find the pattern, and advise the client.
- Pick one C or H task. Collect proof: a before-and-after sample, a checklist you built, or a client note (with permission).
A hypothetical example
This example is made up. A virtual assistant spends most of the week on inbox sorting and calendar booking. AI tools now do much of this. The assistant adds a weekly one-page summary: three decisions that need the client, and three items that can wait. This work needs judgment and trust. The assistant tells the client which tasks the AI handles and which tasks the assistant reviews. The assistant’s value now rests on the review and the advice, not on the sorting.
Start here and go further
Start here: do steps 1 to 3 of the exercise. It takes 20 minutes.
Advanced: repeat the map every quarter. Track how your S tasks move to C. Use the change as proof in your portfolio and in rate talks.
Tradeoffs
- Moving up in value takes time you may not have. Do one small change at a time.
- Not every client wants advice. Some want fast, exact task work. Ask what the client values.
- The “complement” idea is a pattern, not a guarantee.
Do this week
Finish the task map. Choose one S task and write one sentence about how you will move it toward C.
Sources
- Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence, Brynjolfsson, Chandar, and Chen, Stanford Digital Economy Lab, revised August 12, 2026.
- Report on CXAP industry figures presented at the Contact Islands conference in Cebu, by Ehda M. Dagooc, The Freeman via Philstar.com, June 13, 2026.
- Upwork’s In-Demand Skills 2026, Upwork Inc., February 4, 2026.
