Unlock the full potential of AI with Building LLMs for Production—our 470+ page guide to mastering LLMs with practical projects and expert insights!


Stop Calling Automation AI… and the Natural Progression of Intelligent Machines
Latest   Machine Learning

Stop Calling Automation AI… and the Natural Progression of Intelligent Machines

Last Updated on July 20, 2023 by Editorial Team

Author(s): Junis Alico

Originally published on Towards AI.

Defining Automation and AI U+007C Towards AI

Photo credit: Image by Gerd Altmann

In the last few years, I’ve noticed a lot of C-level executives use AI in their keynote speeches and television appearances. They boast about their companies breaking new grounds, advancing technologically. They talk about simplifying processes for themselves and their customers. This is not limited to C-level executives either. A lot of managers are doing it, too. Perhaps it’s all driven by a need to abstract concepts in order to facilitate communication with upper management. It almost seems as if AI is the new buzz word, even though it has been around for quite some time now.

Computer scientists have a better idea: don’t call automation AI! The two are very different and don’t fall under the same category. Automation and AI belong to separate niches altogether.

AI or Simply Automation?

Proper nomenclature has to be followed when referring to either AI or automation. The trend of misinterpreting AI for automation and vice versa has grown uncontrollably mostly due to recent developments in voice recognition systems like Alexa, Google Assistant, and Siri. These technologies automate search processes that until now were heavily manual, and they have automated them to such an extent that this automation is beginning to be perceived as commonplace. To a non-technical person, it may even seem as if the machines are “thinking”. However, this is not the case. Far from it! These products are simply following algorithms, formulas, or pre-programmed pathways. Some pathways are actually so detailed that they cover more than 90% of real-life use cases, giving the impression that technology is “intelligent”.

CEOs clearly don’t want to fall behind their competition. They demand their companies get into AI, the hip new word on the block, without understanding what that truly means, else risk being surpassed by their competitors. The reality is that most of the time companies end up automating a process to simplify it for their end-users. The marketing machine then turns around and tries to sell it as “AI”. This is misleading to both their investors and their customers.

Let’s take a closer look at the differences between automation and AI:

Automation — the ability of a machine (software, hardware, or a combination of the two) to perform tasks without human intervention.

AI — stands for Artificial Intelligence and it aims at building a machine that mimics human behavior, thought processes, and decision making. Some may think of machine learning (ML) as being equivalent to AI. They are not. ML is part of AI. Learning is something that humans do as one of our natural processes.

It’s obvious from the definitions that automation does not necessitate intelligence. Complicated algorithms cannot be classified as intelligent either. AI may contain automation as part of its processes, but these concepts cannot ever be used interchangeably.

In general, AI must involve some sort of network that can make a human-like decision given a problem. These are usually neural networks but don’t have to be. A neural network will have to be constantly trained for it to work. It may even use data as a feedback loop while operating in the real world. Automation, on the other hand, doesn’t need this network for decision making. The decisions are based on specific rules, programmed into the software. The machine will never learn and know how to do anything falling outside of the pre-programmed use cases.

I can clearly remember my Computer Science classes as an undergraduate. Back then, it was all about translating. The goal was to translate paper and Excel artifacts into software-driven processes. The massive amounts of paper processes took a while to undergo the online transformation. It took time to simplify and transform old processes so people could do on a computer the same things they used to do manually. Nowadays, this is no longer the case. Sure, there is the occasional person/team that still uses Excel or paper as their main tools, but the majority relies on online-first methodologies. Everything is done online — storage, backup, searchability, etc. This makes it easier for everyone to work out of one central “location”, accessible from anywhere in the world. Back then, in my early college days, AI was still in its infancy, and neural network libraries were not robust enough to be used to solve real-world problems.

In contrast, nowadays, it’s all about process automation, the same way that back then it was all about process translation. The process automation phase will take much longer than translating paper artifacts. This is natural. During the translation period, the goals were well defined. Sure, there were some improvements introduced along the way. Most of the time, however, the requirements were clear — make the computer allow us to do what we can do on paper. With automation, this is not the case. There are no clear requirements. Automation at its core is about allowing machines to essentially replace humans in repetitive or mundane tasks, improving a process whenever possible. The problem is that no matter how repetitive and well defined a task is, there are always deviations. There’s always some path that requires real-time decision making.

Photo credit: jurvetson on Visualhunt.com / CC BY

Artificial Intelligence is the natural next step in this process. It will allow us to complete our technological journey: from translation to automation to AI. What comes after AI is still to be seen.

Some say that the next big thing is AI and human integration. Projects like Neuralink (also see Elon Musk & team YouTube Nuralink presentation for more information) are gaining traction with recent amazing results. I believe that Neuralink is not the next progression after AI. On the technology evolution timeline, I would place it somewhere in between automation and AI so that the progression would look more like this: translation to automation to Neuralink to AI. Only time will tell if this would be the case.

Current technology is very limiting. It’s not fit to build fully functional intelligent machines. There may be some industry niches where they are using AI to complete tasks, but this is limited to specific use cases (self-driving cars, widget-building robots, speech recognition, etc.). Although still in their infancy, these projects seem to be plagued by problems, requiring constant human intervention and tweaking. AI, the way the general public sees it, is not only non-existent, but I would venture to say not possible with current technologies. We will need paradigm-shifting progress — both in software and hardware — to even consider building such machines. For those thinking that quantum computing is the answer, I don’t agree, especially with already known limitations of such computers. But I do hope that I’m wrong.

So please stop using buzz words like AI to describe anything seemingly “intelligent”. It demeans the field and gives a false impression of having achieved AI already. There’s a lot more research that needs to be done in this field before we can start considering AI for real-world use.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI

Feedback ↓