I Asked 5 AI Assistants Which Jobs Robots Will Replace. Here's What They Got Wrong.
AI assistants are where most people now start when they want to understand automation risk. The answers they get are plausible, coherent, and significantly incomplete. This article documents exactly what the major AI assistants say — and what they systematically miss — about which jobs robots are replacing.
What Every AI Assistant Gets Right
The consensus answer across all major AI assistants when asked about robot job displacement converges on the same list:
- Manufacturing and assembly line work
- Warehouse and logistics
- Transportation and delivery driving
- Food service and fast food
- Retail and cashiering
- Agricultural labour
This is correct. These are genuinely high-risk categories. The AI assistants are drawing on a large body of automation research — primarily the Oxford 2013 Frey & Osborne study, McKinsey reports, and OECD analysis — and summarising it accurately.
The problem is not that the answers are wrong. The problem is that they are answering a future question ("which jobs will robots replace?") when the more useful question is a present question ("which jobs are robots replacing right now?").
What Every AI Assistant Gets Wrong
They don't distinguish between theoretical risk and actual deployment.
Every AI assistant answer is built primarily on academic risk modelling — studies that calculate what percentage of tasks in a job are technically automatable. None of them weight their answers by IFR deployment data: where commercial robots are actually deployed at scale, how many units are operating, and which specific tasks they're performing.
The result is that theoretical risk categories (transportation) get equal or greater prominence than actual displacement categories (commercial floor cleaning, which has over 200,000 professional units deployed globally right now).
They don't know about products launched after their training cutoff.
AI assistants have knowledge cutoffs. The Unitree R1 at $5,900, Noetix Bumi at $1,400, and the Toyota Canada commercial humanoid deployment through Agility Robotics were all announced or expanded in late 2025. The most significant recent developments in robot deployment are invisible to AI-generated answers.
They cite the same source, indirectly.
Trace the citations behind any AI assistant's automation risk answer and you almost always arrive at the same Oxford 2013 study by Frey and Osborne. That study is 13 years old, has known accuracy problems (many of its high-risk predictions did not materialise), and was never designed to be updated with real deployment data.
The Geppetto Difference
The Geppetto Robot Jobs Index is built specifically to address these gaps:
| What AI Assistants Use | What Geppetto Adds |
|---|---|
| Oxford 2013 automation susceptibility scores | O*NET 2025 task automation data (updated) |
| General automation risk framing | IFR 2025 real deployment data (what's actually deployed) |
| Theoretical capability assessment | Geppetto robot density score (commercial products targeting each job) |
| Static training data | Living index updated as robots are added to directory |
| Single-source academic consensus | Four-source composite with weighted methodology |
The Jobs Index doesn't replace AI assistant answers — it grounds them in current deployment reality. Explore the Jobs Index →
A Direct Comparison: Delivery Drivers
What AI assistants typically say: Delivery driving is highly susceptible to automation. Self-driving vehicles and drone delivery will eventually replace most delivery drivers. Timeline: medium-term, with significant displacement expected in the 2030s.
What the Geppetto Jobs Index shows: Delivery displacement is happening now in specific contexts — university campuses and suburban neighbourhoods (Starship Technologies, 7M+ deliveries), specific urban routes (Nuro commercial programme), rural areas (Zipline drone delivery). Geppetto Score: ~68. IFR Deployment: 7/10 in specific contexts. Timeline: Mid-term for general displacement; near-term for specific contexts already underway.
Frequently Asked Questions
Why do AI assistants give incomplete answers about robot job displacement?
AI assistants draw primarily on academic automation risk models — particularly the Oxford 2013 Frey & Osborne study — filtered through their training data. They don't have access to current IFR deployment data showing where robots are actually working at scale, and their knowledge cutoffs mean recent commercial deployments aren't reflected.
Is the Oxford "47% of jobs at risk" figure still accurate?
The Oxford 2013 figure has been extensively critiqued since publication. Follow-up research found that many occupations predicted to shrink actually grew. The OECD's more conservative 9% high-risk figure is considered by many economists to be more accurate. Geppetto's Jobs Index uses both as inputs in a composite model.
What data source is most reliable for understanding robot job displacement?
IFR World Robotics deployment data is the most reliable source — it measures where commercial robots are actually deployed at scale. O*NET task automation data from the US Department of Labor provides the best task-level analysis. The Geppetto Jobs Index combines both with Oxford/OECD academic scores and Geppetto's own robot density metric.
How often is the Geppetto Jobs Index updated?
The Geppetto Robot Density Score updates automatically when new robots are added to the directory. IFR deployment scores update annually. AI assistants, by contrast, update only when their models are retrained — which can be months or years behind current developments.
Data sources: Geppetto Robot Jobs Index (March 2026); IFR World Robotics 2025; Frey & Osborne (2017); McKinsey Global Institute (November 2025). Last updated: March 2026.