GB710 Unit 4 Historical Parallel Analysis Example

Machines Were Ending Work in 1964 Too: A Unit 4 Historical Parallel Analysis of AI and Graduate Hiring

[Student Name]

Purdue University Global

GB710: Contemporary Challenges in Business

Unit 4 Historical Parallel Analysis

[Instructor Name]

[Date]

The firm is composite and illustrative; the historical episodes, reports and studies cited are real.

The Present Pattern, Dated

Generative AI tools became widely available in late 2022, and accounting software vendors added them to audit and tax products in 2023 and 2024. The candidate's composite firm adopted two such tools in early 2024 and cut its fall graduate intake by about a third later that year. Payroll records analyzed in a 2025 working paper show the youngest workers in exposed occupations losing ground against comparison groups from late 2022 (Brynjolfsson et al., 2025). The pattern to be explained is therefore a shrinking entry point in occupations where new tools can perform much of the routine work beginners used to do. Three features define it. The tools perform cognitive work, not physical or clerical work. They are adopted quickly, within a year or two of release rather than over a decade. And the work they perform is the work that firms have traditionally used to train new professionals. A useful parallel should share at least one of these features, and the analysis notes which features each episode shares.

Parallel 1: The Automation Commission of the 1960s

In 1964, amid fears that computers and automated machinery would cause mass unemployment, Congress created the National Commission on Technology, Automation, and Economic Progress. Its report, issued in 1966, concluded that technology eliminates jobs, not work: specific jobs disappear, but total demand for labor depends on economic growth, and new kinds of work appear (National Commission on Technology, Automation, and Economic Progress, 1966). Employment grew strongly through the following decade, and unemployment fell to its lowest levels in years by the late 1960s. The commission's fears, which had seemed reasonable given the pace of computerization in factories and offices, did not come true at the level of the whole economy, though particular occupations and regions did suffer.

What transfers. The distinction between jobs and work applies directly. AI may eliminate specific tasks without reducing the total work an accounting firm needs done, if the firm's clients demand more analysis as routine processing gets cheaper.

What does not transfer. The commission was concerned with the whole economy's employment level. The firm's question is narrower: whether its own entry-level positions, a specific rung on a specific ladder, will shrink. A growing economy can coexist with a missing rung.

Parallel 2: Bank Tellers and ATMs

Automated teller machines spread through American banks from the 1970s onward, and many expected teller jobs to disappear. Instead, as Bessen (2015) documents, the number of tellers held up for decades. ATMs reduced the number of tellers needed per branch, which made branches cheaper to operate, so banks opened more of them. Tellers' work shifted from handling cash toward customer relationships and sales. Only later, with online and mobile banking, did teller employment decline.

What transfers. The mechanism: automation of routine tasks lowered the cost of each unit of service, which increased demand for the service and shifted the remaining human work toward judgment and relationships. If AI makes routine accounting work cheaper, firms might serve more clients or offer more advisory work, increasing demand for staff who can do the non-routine parts.

What does not transfer. Tellers' new work was still entry-level work. In accounting, the work AI takes over is the work that trains beginners to become reviewers. If the rung shifts upward, beginners may lack the routine work through which they used to learn.

Parallel 3: Spreadsheets and Bookkeeping

Electronic spreadsheets arrived in the late 1970s and 1980s and automated much of the calculation that bookkeepers and accounting clerks performed by hand. Employment of bookkeeping clerks declined over the following decades, while employment of accountants and financial analysts grew. Autor et al. (2003) show that computerization reduced demand for routine cognitive tasks while increasing demand for non-routine analytical tasks, changing the mix of occupations rather than the total.

What transfers. The rung moved. Work that had been done by clerks with little training was absorbed by software, and the entry point to the profession moved toward analytical work requiring more education.

What does not transfer. The clerks displaced by spreadsheets were largely a different group from the graduates who became accountants. Today's displacement falls on the graduates themselves.

Each parallel suggests the work will change rather than vanish; none promises the first rung stays where beginners can reach it.

What the Parallels Suggest Together

The three episodes share a pattern: automation of routine tasks changed which tasks people did and moved the entry point of occupations, while total employment followed broader economic conditions. They suggest three hypotheses for the present case. First, AI may shift first-year accounting work toward review, client contact and judgment, rather than eliminate it. Second, the firms that redesign first-year work may gain an advantage in developing future reviewers. Third, if most firms simply cut intake, the profession could face a shortage of experienced staff several years later, as the missing first-years fail to become managers.

Why Parallels Persuade Partners

Historical parallels appear often in the firm's own discussions, usually as reassurance: accountants survived spreadsheets, so they will survive AI. The appeal is understandable. A parallel offers a story with a known ending, which feels more solid than a forecast. But a parallel persuades only if the mechanism that produced the earlier outcome is present now. Partners who cite the spreadsheet example rarely ask who bore its costs, and the answer, clerks rather than accountants, is precisely the feature that makes the present case different. The analysis therefore asks of each parallel not whether it ended well, but for whom, and through what mechanism.

What the Parallels Cannot Show

None of the episodes involved a technology that performs the same tasks beginners use to learn. That feature may make the current case different. The analysis therefore uses the parallels to generate questions for the comparative case study in Unit 5 rather than to predict an outcome. The most important of those questions is whether firms that keep hiring graduates, and redesign what they do, develop reviewers faster than firms that cut intake. A second is whether clients notice any difference in the quality or speed of work. A third is how long it takes before the effects of a thinner entry class show up in a firm's management ranks, since the parallels suggest such effects arrive years after the decision that caused them.

References

Autor, D. H., Levy, F., & Murnane, R. J. (2003). The skill content of recent technological change: An empirical exploration. The Quarterly Journal of Economics, 118(4), 1279-1333. https://doi.org/10.1162/003355303322552801

Bessen, J. (2015). Learning by doing: The real connection between innovation, wages, and wealth. Yale University Press.

Brynjolfsson, E., Chandar, B., & Chen, R. (2025). Canaries in the coal mine? Six facts about the recent employment effects of artificial intelligence [Working paper]. Stanford Digital Economy Lab.

National Commission on Technology, Automation, and Economic Progress. (1966). Technology and the American economy. U.S. Government Printing Office.