Small Businesses Bolting AI Onto Old Workflows Are Repeating an 1890s Factory Mistake, Post Claims

A post comparing the AI rollout in small businesses to a mistake factory owners made more than a century ago has resurfaced a 1990 academic paper on why transformative technology often fails to boost productivity right away. The post, published by X account @HowToPrompt_, argues that 'we're making the exact same multi-billion-dollar mistake they made in the 1890s' by adopting artificial intelligence without changing how businesses actually operate.

The Viral Claim That Caught Our Attention

The post points to the shift from steam power to electricity in late nineteenth century factories. Steam engines ran on a single central power source, distributed through a network of belts and shafts to every machine on the floor.

When electric dynamos arrived, the post states, factory owners simply swapped the engine. 'Factory owners bought the new electric motors, ripped out the steam engines, and plugged the electricity straight into the exact same chaotic belt systems,' it reads. The result, according to the post, was that 'productivity barely moved... it actually took 30 years for the real economic boom to happen,' because businesses 'didn't redesign the factory. they just swapped the engine.'

The post then applies the same logic to AI adoption today, stating that 'small business owners are buying ChatGPT subscriptions, slapping an AI chatbot onto a legacy software stack, and wondering why their overall efficiency and profits haven't skyrocketed.'

What a 1990 Academic Paper Actually Found

The historical claim in the post is drawn from a real academic paper. Economist Paul A. David, then of Stanford University, published 'The Dynamo and the Computer: An Historical Perspective on the Modern Productivity Paradox' in the American Economic Review in May 1990.

David's paper argues that factories initially kept their old centralised layout even after installing electric dynamos, and that the real efficiency gains only arrived once businesses gave each machine its own motor and rebuilt the assembly line around it. Researchers have traced this shift to the 1920s, decades after electricity first became available, with David's work remaining central to that historical analysis.

David's paper remains widely cited by other researchers, and interest in his work grew again following his death in 2025, at age 87.

Why AI Adoption May Be a 30-Year Bet

The post's argument is that the same pattern is repeating with AI. It states that 'you can't just drop an LLM into a broken, manual corporate process and expect it to print money,' and that meaningful gains require businesses to 'completely redesign your operations around agentic workflows' rather than adding a chatbot on top of existing systems.

For small businesses specifically, this suggests that subscription costs for AI tools may not translate into visible returns unless staff processes, software integrations and decision-making structures are rebuilt around the technology, not simply layered with it.

The post summarises this directly, stating: 'stop building steam factories with electric motors.'

The comparison lands at a moment when small businesses are under pressure to adopt AI quickly or risk falling behind competitors, yet many report limited measurable returns on that investment so far. David's research suggests that gap may not be a sign AI does not work, but a sign that meaningful productivity gains from major new technology historically take years, and require businesses to rethink how they operate rather than simply installing new tools into old routines.