Replacing humans with AI is not going to work today!

A few weeks ago, I wrote the article “Replacing humans with AI today is a recipe for failure”, in which I explored how the rush to automate and change/substitute human skill with artificial intelligence is producing more harm than good in business and service delivery.

Building on that foundation, I would like to focus this week on why the effort to replace humans with AI is going terribly wrong. I will share some of the flawed implementations and misplaced expectations that led to breakdowns in customer experience, ethical challenges and hard lessons from companies that learnt the limits of AI the hard way. This should probably reduce the pressure or urge of the many companies and governments that wish to jump on the AI bandwagon that promises cost savings, 24/7 operations and higher productivity — apart from promoting the notorious AI-ready badge.

In the last 2 years, virtually every leader of an institution had the goal to adopt AI in one way or another. Nevertheless, harsh reality was the outcome, especially with those that simply replaced humans with AI systems yet failed in a big and disastrous way.

Not everyone likes to share their failure, but the proof is in the pudding on the mere/small number of initiatives that actually succeeded. In Oman — and I know for a fact — there have been many seminars, workshops and even startups that are working on AI solutions, yet these all remain to be a showcase for a better future. On paper, not in practice though!

A report by one of the leading technology and science institutes in the world, Massachusetts Institute of Technology, found that only 5–7% of AI projects generate real good value (and immediate financial revenue), yet 93–95% of implementations fall short of expectations.

Furthermore, over 55% of companies that replaced employees with AI now regret it. Clearly, replacing humans with AI today is neither healthy nor wealthy for any organisation. It’s too early for ambitious initiatives but definitely is improving day by day. AI today does wonders. Yet if one doesn’t adopt it slowly and/or gradually, their fate is most probably doom. The MIT report, which is available readily online to read, is a revelation and indeed an eye-opener!

One of the most striking examples of AI gone wrong comes from the largest fast-food chain restaurant, McDonald's, which introduced an AI voice-ordering system and installed it in over 100 locations. The system was meant to streamline orders and reduce wait times; nevertheless, the outcome was a disaster. The AI system misunderstood customer requests and ultimately caused both confusion (eg, wrong orders). McDonald's terminated the initiative to save the company from reputational and financial impact. I did share in my previous article another example of the Swedish fintech giant company called Klarna, which made headlines after laying off around 700 customer-service employees and replacing much of the function with AI systems. The result was a catastrophe.

So what is the problem? Why don’t AI replacement and adoption really work? I will tell you, assuming technology can think the way humans do is the main problem. At least today, they can’t and I can’t promise nor predict if this won’t happen/change in the future. AI is great at patterns and repetition, but it falls apart when put into a situation a real smart human can do and at various disciplines.

Conversations sometimes get illusional and/or off-script too, customers get emotional and situations at many times require judgement rather than logic. Rushing into adopting AI without rethinking how work actually happens is a big mistake. Humans continue to add value where AI cannot due to being a machine in the first place.

To conclude, let me reemphasise that I am an AI supporter myself; however, I still believe that AI is not ready to replace quality/educated humans. AI isn’t failing, but the way we’re using it is. Judgement, empathy and control are something that only humans can lead today, not machines and certainly not AI. The real opportunity lies in collaboration, not substitution and those who understand this will get far more value from AI than those who learn it the hard way. Until we catch up again next week, believe in yourself; you are not replaceable yet!