Why careless adoption of AI backfires so easily

Artificial intelligence (AI) is rapidly becoming commonplace, despite statistics showing^[1] that only approximately 7% to 13% (depending on size) of companies have incorporated AI into their regular business workflows.

Adoption in specific business functions is far higher, with up to 78% of companies reporting use of AI tools^[2] in at least one business area. And more than 90% of companies plan to increase AI investment within three years^[3].

This surge in adoption is underpinned by expectations of significant efficiency gains and cost reduction^[4].

Widespread implementation of AI is also accompanied by layoffs^[5]. Estimates vary, but it’s clear that within the next decade, millions of jobs^[6] will be reshaped or even replaced thanks to AI.

However, despite the lofty promises of AI, many companies aren’t seeing the payoff. Data on productivity gains from AI use is murky at best^[7], and many companies are facing costly implementation failures^[8].

Organisations are falling for what is known as the doorman fallacy: reducing rich and complex human roles to a single task and replacing people with AI. This overlooks the nuanced interactions and adaptability humans bring to their work.

What is the doorman fallacy?

British advertising executive Rory Sutherland^[9] coined the term “doorman fallacy” in his 2019 book Alchemy. Sutherland uses the concept of the humble hotel doorman to illustrate how businesses can misjudge the value a person brings to the role.

To a business consultant, a doorman appears to simply stand by the entrance. They engage in small talk with those coming and going, and occasionally operate the door.

If that’s the entirety of the job, a technological solution can easily replace the doorman, reducing costs. However, this strips away the true complexity of what a doorman provides.

The role is multifaceted, with intangible functions that extend beyond just handling the door. Doormen help guests feel welcome, hail taxis, enhance security, discourage unwelcome behaviour, and offer personalised attention to regulars. Even the mere presence of a doorman elevates the prestige of a hotel or residence, boosting guests’ perception of quality.

When you ignore all these intangible benefits, it’s easy to argue the role can be automated. This is the doorman fallacy – removing a human role because technology can imitate its simplest function, while ignoring the layers of nuance, service and human presence that give the role its true value.

Doormen everywhere

As AI becomes increasingly common, many companies have started evaluating employees the way a consultant might evaluate a doorman. The judgement is based purely on the most visible, basic tasks they perform, such as taking food orders or answering phones.

The focus is on what can be automated and what costs can be reduced. What often gets overlooked is the broader value a person brings through context, judgement, and the countless invisible contributions that support a thriving workplace.

This narrow view leads straight into the doorman fallacy, assuming a role is simple because only the obvious parts are seen.

Earlier this year, the Commonwealth Bank of Australia fired 45 customer service staff^[10] and rolled out an AI voice bot, claiming the bot drastically cut call volumes.

After the workers’ union challenged the layoffs, the bank reversed its decision, admitting it “did not adequately consider all relevant business considerations and this error meant the roles were not redundant”.

In the United States, fast-food chain Taco Bell has been rolling out voice AI^[11] in its drive-throughs since last year, in hopes of cutting errors and speeding up service.

After a barrage of customer complaints and social media videos documenting^[12] various glitches, the company is now rethinking its AI use. Taco Bell’s chief technology officer conceded to the Wall Street Journal it might not make sense^[13] to only use AI at drive-through and that human staff might handle things better, especially during busy times.

These are not isolated examples. A recent report from software platform Orgvue states up to 55% of the companies^[14] that replaced employees with AI now acknowledge they moved too quickly. Some companies are rehiring the very people they let go^[15].

On top of that, consumers dislike dealing with AI^[16] in customer service settings, and most say they’d likely choose a competitor that doesn’t use AI^[17].

A job is more than a list of tasks

To avoid the doorman fallacy, companies must recognise jobs are more than the visible tasks listed on a job description.

Employees frequently contribute in subtle ways that leaders don’t see day-to-day, yet those contributions hold real value for customers and organisations as a whole.

Smart AI adoption requires a full understanding of the human elements inside every role. The concept of “efficiency” should be expanded to valuing customer experience and long-term outcomes as much as cost savings.

Before a company attempts to automate any roles and hand tasks over to AI, it must have a deep understanding of the roles in question. If the task needs human oversight and intervention, it’s not a good candidate for automation.

AI can be implemented in roles that don’t require human oversight, such as data entry, image processing, or even predictive maintenance^[18] that monitors the health of equipment – roles that are rule based and clearly measurable, freeing up people to do other things.

The evidence so far is clear: the best way to use AI is to pair it with human judgement. This approach preserves the parts of work where context, personal touch, and trust matter.

By supplementing human roles with AI, standardised and repetitive tasks can be completed efficiently, allowing individuals to focus on contextual work where a human touch is important.

References

1. ^{^} statistics showing (www.goldmansachs.com)
2. ^{^} 78% of companies reporting use of AI tools (www.mckinsey.com)
3. ^{^} plan to increase AI investment within three years (www.mckinsey.com)
4. ^{^} significant efficiency gains and cost reduction (www.bcg.com)
5. ^{^} accompanied by layoffs (www.cnbc.com)
6. ^{^} millions of jobs (www.abc.net.au)
7. ^{^} from AI use is murky at best (theconversation.com)
8. ^{^} costly implementation failures (gizmodo.com)
9. ^{^} Rory Sutherland (en.wikipedia.org)
10. ^{^} fired 45 customer service staff (gizmodo.com)
11. ^{^} rolling out voice AI (omilia.com)
12. ^{^} social media videos documenting (www.tiktok.com)
13. ^{^} it might not make sense (www.wsj.com)
14. ^{^} 55% of the companies (www.orgvue.com)
15. ^{^} rehiring the very people they let go (ia.acs.org.au)
16. ^{^} dislike dealing with AI (kinsta.com)
17. ^{^} competitor that doesn’t use AI (www.techspot.com)
18. ^{^} predictive maintenance (www.ibm.com)