I admit, I like flying. It is good I do, because last year I spent a large part of my time in the air. Flying allows me to be alone with my laptop. For those fortunate enough to live in a world where laptops have a value by themselves, and not only based on their capacity to connect with someone else, sitting without the annoying stimulation produced by new email notifications, social network messages, offers (I think I make my point..), is just a valuable opportunity to put my thoughts in order.
AI is probably the latest term of a long series of notorious jargon abbreviations the IT community has gifted the world. For a significant portion of the population, another step towards a time where machines will do everything and humans will just wait for their drinks seated at the edge of the swimming pool, served by a bionic device, probably manufactured in Japan or China. Not a surprising view, considering the Hollywood contribution to western civilization. For others, it just provides a catastrophic view of the future, where machines rule because of human arrogance in playing God. Do not dismiss other, equally uninformed, voice contributions here.
For a good few decision makers in the IT and Telecom industries, it just represents a new way to do the same old thing. In other words, a fresh technical term used to keep charging customers. There is no need to understand what it is about, or the value it can provide to the market and, ultimately, to those customers. It is okay adding "powered by AI" at least 6 times in the customer deck and naming it as many times during the presentation. Of course, the more substantial questions qualify as "technical" and are diverted to the "expert" (details are boring…).
All the above have contributed to creating the AI bubble. But listen, bubbles are not that bad. They generate significant economic activity, higher salaries for professionals, big demand for skills. I had 698,747 Data Scientists in my 1st, 2nd and 3rd linked-in connections last time I checked. A notable figure, considering advanced statistics was a subject almost everyone I knew at university tried to avoid.
IT and Telecoms have played a central role in global GDP growth over the last 35 years, considering the entire industry was virtually non-existent 50 years ago, it is reasonable to think the existence of bubbles is the price to pay for such growth. Bubbles come and go, but they always leave something behind. For instance, the world as we know it today, would not be the same without the dot-com bubble and all the technology created to push the internet into a global, gigantic marketplace.
The reality is that despite the general beliefs and hopes (or fears), I do not see a Japanese built robot preparing my salad (not yet at least). During fifteen years in data science projects, in both industry and academia, I constantly saw the setting of unrealistic expectations about the value of the current state of "intelligent algorithms" and what they can, and cannot do, for enterprise. However, the real question is why? Why companies invest so aggressively in AI, and it seems only Google and Amazon are getting amazing results? It is true, they do get amazing results, but it is also true they invest a few times more than the average.
The possibility of creating intelligence artificially is captivating and fascinating, but what is intelligence? Having a definition would be helpful, before trying to re-create it artificially. Generally, this is the point at which people start to feel uncomfortable with the conversation. Are intelligence and knowledge the same thing? Is it possible to define a behaviour (human or not) as intelligent because it shows an in-depth knowledge in a specific delimited field? Can we define as intelligent an algorithm because it can identify a human face on a picture, or a tumour on an X-ray? If an algorithm can process and correlate millions of labelled X-rays, and blood tests in a few hours, making accurate predictions about the progression of a disease, is it safe to imply that algorithms are smarter than consultants? Perhaps it is necessary to be very smart, and draw a line between intelligent and repetitive behaviours (regardless of their complexity), which not only applies to machines.
Maybe the best thing about the AI bubble, is it forces us to answer, or at least to consider, extra-intelligent questions. Is it not a great discipline that puts philosophers and computer scientists in the same arena? Maybe that is why the old models do not fly commercially. Maybe that is why we are witnessing the birth of a new era, like electricity or the creation of the first electronic computer. Maybe, and paradoxically, the tireless and relentless need of humankind to understand and create intelligent things, will end up generating more intelligent humans.
As mentioned, bubbles are not that bad, except it is best not to be around when they burst.