Why Artificial Intelligence and Robotics do not have to be connected

In a recent article I wrote about “Robotics" (also known as "Robotic Process Automation" RPA) and the different opportunities and challenges they bring. This created some questions whether Robots relate to Artificial Intelligence (AI) and to what extend.

 

Software Robots are next generation macro computer programs that can perform more complex tasks. The indention of a robot is to replace a specific administrative task a human had previously performed. What does that mean?

Many academical and rigorous papers were already written about potential differences (Robots versus Machine Learning versus Artificial Intelligence). And if you talk to specialised solution providers you often get piles of PowerPoints selling taxonomies, roadmaps, product layers and AI frameworks.


“Artificial Intelligence can help Software Robots
to become more effective in the future. 
But not all Robots need Artificial Intelligence
to perform their tasks"



I don’t intend to challenge those valuable papers. Some are very complex and leave the reader a bit in the dark. If you allow yourself to look at it from a pragmatic approach the difference becomes quite apparent.
  • Artificial Intelligence describes the principle of machines and computers to perform tasks we would mark as “intelligent” while self-learning at the same time
  • The idea is not new and has been visioned since long
  • During the genesis of computer development, scientists mainly focused on making computers faster in executing repetitive standard tasks. This was a natural focus as computer power was minimal compared to today
  • The ever increasing computer power during the last decades allowed new technology architectures to be built and hence, enabled computers to perform more complex tasks
  • In addition, we understand much better now how concluding, learning and decision making works. And more important, how to teach computers to learn for themselves and therefore, modify their future problem solving approach
  • Artificial Intelligence is therefore able to go much further. Examples are understanding what somebody is saying (Sound Recognition), recognising faces in photos and videos (Vision Recognition), turning data into actionable information (Context Recognition), understanding the meaning of written text in Email or Chat (Language Processing), predicting purchase behaviours of customers (Predictive Analysis) or understanding why something is happening (Interpretation Formulation)

The thorough self-learning part is really what separates a Robot from an AI system. Robots can perform complex tasks. Robots have also the capabilities to adapt and fine-tune their behaviours which makes them appear smart. Artificial Intelligence systems on other hand are capable of learning for themselves. And this will open up opportunities we might even not be able to imagine yet. 

 

AI is more an evolution that can support many use-cases including those of robots. It’s a bit like a child that grows to a teenager and then to an adult. It’s an evolution that includes learning, intelligent behaviour and more complex action taking over time.

 

Artificial Intelligence will help many software robots to become even more effective in the future. But not all robots need Artificial Inteligence to perform their tasks. 

 

Chris Frey @chrisfrey.com