TL;DR
ATM growth initially reduced the number of tellers needed at each U.S. branch, while lower operating costs and branch expansion helped teller employment rise modestly through 2010. Later, as mobile banking became the main channel for more customers and branch networks contracted, teller employment fell in the United States and branches declined sharply in Germany. The source uses this history to frame a broader question about AI and banking, but gives no data on current AI deployments or their effects on jobs.
The episode centers on the automated teller machine, introduced at a Barclays branch in Enfield, north London, on June 27, 1967. Actor Reg Varney demonstrated the machine, which dispensed up to ten pounds in one-pound notes. Early customers used a special paper cheque marked with carbon-14 and entered a personal code; plastic bank cards had not yet arrived. The machine handled cash transactions that had been part of a teller’s work, while branches still operated during the same period.
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
ATM numbers in the United States grew from about 100,000 in 1995 to about 400,000 in 2010, roughly quadrupling, according to figures from the Bank for International Settlements cited in economist James Bessen’s analysis. Yet U.S. teller employment increased modestly over the longer period from about 500,000 in 1980 to about 550,000 in 2010, as David Autor summarized Bessen’s findings in 2015. These figures cover different time spans and do not show that ATMs alone caused employment to rise.
Bessen’s explanation, as presented in the episode, is that branches needed fewer tellers but became cheaper to operate. In a typical U.S. urban market, teller staffing per branch fell from about 20 to 13 between 1988 and 2004. Over that same period, the number of branches in U.S. cities grew by 43%. The FDIC’s national count of commercial bank offices, including head offices and branches, rose from 58,937 in 1988 to 77,302 in 2004 and peaked at 88,522 in 2012.
Exploring the AI Paradox in Banking Services
The ATM era reshaped teller work over decades. Its history raises questions about AI, but the evidence here does not measure AI’s effects.
ATMs changed branch work
Fewer tellers were needed at each branch, while lower operating costs and branch expansion helped U.S. teller employment rise modestly through 2010.
- Typical urban teller staffing fell from about 20 to 13.
- U.S. city branches grew by 43%.
- Teller employment rose from about 500,000 to 550,000.
- ATM numbers grew from about 100,000 in 1995 to about 400,000 in 2010.
Banking shifted toward mobile
Later, mobile banking became the main channel for more customers as teller employment and branch networks declined.
ATMs took on cash transactions while tellers shifted toward customer advice and sales.
By 2023, more U.S. households mainly used mobile banking than a teller.
U.S. teller employment fell after 2008, and German branch counts declined by 2025.
What this says about AI
The history offers a comparison, not a forecast: the material gives no direct measurements of AI use or its effects on bank jobs, services, or costs.
| Evidence in the account | What remains unknown |
|---|---|
| ATM, mobile banking, teller employment, and branch figures | Which AI systems banks use and what outcomes they produce |
| Changes across different periods and measures | Whether AI will follow the ATM era’s path |
| Later job and branch declines coincide with wider channel shifts | How much any one technology caused those changes |
How Branch Economics Changed Teller Work
The figures complicate a simple prediction that machines performing a job’s core tasks will quickly eliminate the workers who perform it. In this account, ATMs reduced the staffing needed at each branch, while lower running costs and branch expansion supported more locations. The result was a change in where teller work happened and what tellers did, rather than an immediate disappearance of the occupation.
The later numbers point to a different period. In 2023, 48.3% of U.S. households with a bank account mainly used mobile banking, while 15.1% mainly used a teller, according to the source. U.S. teller employment fell from about 600,500 in 2008 to 339,200 in 2025, using the Bureau of Labor Statistics count cited there. In Germany, bank branches declined from 45,467 at the end of 2004 to 16,799 at the end of 2025. Those trends matter to customers and workers because they describe a shift toward remote banking and fewer in-person locations. The figures alone do not establish that mobile banking or any single technology caused the job and branch changes.
The episode’s relevance to AI is its historical comparison: the labor effects of automation can depend on business decisions and customer behavior as well as on what a machine can do. However, the material provides no direct measurements of AI’s current effect on bank jobs, services, or costs. The ATM example is context for that question, not evidence that AI will follow the same path.
Products Worth Considering

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)

DJI Mini 4K Fly More Combo Drone with 4K Camera and 3 Batteries

Jackery HomePower 3000 Portable Power Station, 3072Wh
As an affiliate, we earn on qualifying purchases.
From Enfield to Mobile Banking
The Barclays demonstration was followed by early cash machines in other countries. On May 27, 1968, the Kreissparkasse Tübingen opened what the source identifies as Germany’s first cash machine, built by AEG-Telefunken and Ostertag; customers could withdraw up to 400 marks. In September 1969, Chemical Bank installed a Docutel cash dispenser in Rockville Centre, New York, according to the Smithsonian and the Computer History Museum.
In the United States, ATM growth accelerated from the mid-1990s. The source describes tellers shifting toward customer advice and sales as machines took on routine cash handling. It also notes that deregulation made it easier for banks to open branches, another factor in the expansion Bessen examined. The FDIC office count later fell to 72,658 in 2025, below its 2012 peak.
The article then moves from ATMs to mobile banking as the later change in how customers use banks. Its figures also point to cash machines retreating in Germany, China, and Japan, but provide no counts or dates for those changes. This history tracks several distinct developments over decades; it does not isolate the effect of any single technology.
What the Figures Do Not Show
The source does not provide a direct study of AI use in banking, identify particular AI systems used by banks, or quantify AI-related job changes. Its title raises an AI paradox, but the evidence presented concerns ATMs, mobile banking, teller employment, and branch counts. It is therefore unclear from this material how AI is affecting banking services or whether its labor effects will resemble those associated with ATMs.
The employment and branch figures also come from different measures and periods. The episode attributes the U.S. teller count for 2008–2025 to the Bureau of Labor Statistics and the longer 1980–2010 comparison to Bessen’s analysis as summarized by Autor. The data do not establish why teller jobs fell after 2008, nor do they isolate the effects of mobile banking, other technology, bank consolidation, or changes in customer demand. The source gives no detailed figures for cash-machine closures in Germany, China, or Japan.
The Next Test for Banking Automation
The next question raised by the episode is whether banks will use AI to change branch operations, customer service, or staffing, and how those changes will compare with the ATM era. The source names no upcoming study, policy decision, or industry announcement that would resolve this. Readers looking for an answer will need dated, bank-specific evidence on AI deployments, service outcomes, costs, and employment, alongside clear comparisons with earlier periods.
For now, the historical account ends with a measured finding: ATMs reduced teller needs per branch while U.S. teller employment rose modestly through 2010; later, teller employment declined as mobile banking became common and branches contracted. Whether AI produces a similar sequence remains an open question.
Key Questions
Did ATMs immediately eliminate U.S. bank teller jobs?
No. The figures cited in the source show U.S. teller employment rising modestly from about 500,000 in 1980 to about 550,000 in 2010, even as ATM numbers roughly quadrupled from 1995 to 2010.
Why did teller employment rise while ATMs spread?
James Bessen’s explanation, as presented in the episode, is that branches needed fewer tellers but cost less to operate, and banks opened more branches. The account also identifies deregulation as a factor that made branch expansion easier.
What happened to U.S. teller employment after 2008?
The source cites a Bureau of Labor Statistics count showing teller jobs falling from about 600,500 in 2008 to 339,200 in 2025. It does not establish a single cause for that decline.
Does the episode show that AI is reducing bank jobs?
No. It discusses earlier automation and later mobile banking but provides no direct data on AI adoption or AI-related employment changes in banking.
Source: Thorsten Meyer AI
