Unless you have been living under a rock, you must have seen the frenzy around AI. Companies and executives compete with one another to see who mentions AI the most in every communication. Any hint that a firm is slow to leverage AI and lagging behind competitors will result in lower stock prices and scrutinty from investors. This gold-rush mentality leads to some executives trying HARD to find a use case for AI in their business where there is virtually not. Take what Domino’s Pizza, which just makes and delivers pizzas, hilariously had this to say about how they incorporate AI

While some of the progress and products out there look impressive and exciting, I think AI is a bit overhyped in the short term.

What AI is really good at is to compile information and think/act linearly at the speed that no human can accomplish. Want to find information on a fairly complex topic? Give the likes of ChatGPT a prompt! Want to know how to get do something in Python? Just ask Gemini! Looking to crunch some numbers and create a regression model? Again, hit an AI-powered chatbot.

I usually tend to be skeptical of a new thing. In the past few years, we had Metaverse, VR, crypto, electric vehicles (EVs) and NFT become massively popular almost overnight and wither just as fast. I don’t believe AI is as impractical as some of the other hyped technologies. There will be a place for AI in our society, like there has always been. We’ll unlock more efficiency in both daily and corporate lives with AI.

But I doubt the hyperbole of how AI will change the world and I suspect that there is a big gap between the soundbites from executives and what end users really find helpful.

First, efficiency is not always the best thing to do. AI can offer a full code if prompted, but in that case, what does the prompter get to learn? Every organization or assignment has its unique elements that no chatbot can replicate. Yes, AI can provide a template, but one must understand what the template does to make the appropriate adaptation. And the only way to understand code is to code. Not to prompt.

AI now can offer a summary of books or earnings call transcripts. But you read books or transcripts to internalize learnings and observations. What AI provides can save you time, but will deprive you of an opportunity to add to your subconscious learning.

The same goes for writing. Yes, AI can write a much better post in less time than what I am typing here. But it wouldn’t make me a better communicator or a writor. The only way for me to communicate and write better is to write it myself. To express my personality and ideas in my own words. Even grammatical or vocabulary errors are my own and that’s how it should be.

The second reason why I doubt that AI will replace us is that I don’t believe it can reason as humans. We possess a unique ability to put numbers in context and tell stories. Sure, AI can do maths emphatically faster than us, but someone needs to translate all numbers into learnings and actions. Can AI do that? Will AI ever do that? Around 2012 or 12 years ago, we heard tech executives predict that we would have self-driving cars on the road in 5-7 years. To this day, we still don’t. Self-driving cars are a form of artificial intelligence. It’s not driving straight or turning corners that is the problem. It’s handling ethically tricky situations such as the trolley problem that makes it exceedingly difficult. Even if we can get over the technical challenge, who is to decide what self-driving cars should do?

The same can be said about searches on the Internet. Yes, ChatGPT can give you answers and save you from browsing different websites. But is that what you want? When we research on a topic, we want different perspectives, pros & cons, and we want to vet the sources. Can ChatGPT or its peers do that? Even that is the case, who gets to decide what the AI-powered chatbots show us? Isn’t it some form of censorship?

In short, I am skeptically bullish on AI. It can help us achieve the low-hanging fruits in terms of efficiency. It is entirely possible that AI may push the boundaries of what we can do and how it can change our lives. The problem is that no-one really has a crystal ball to say confidently what it looks like. In the famous adoption S-curve, where would you say we are now? Are we closer to Solution Search or Market Expansion?

Not until there is more evidence, I believe it’s overhyped. At least in the short term.