What is AI? It’s the most amazing and bizarre human-created tool. A tool that create other tools and even improve itself. That’s the greatest invention ever. By itself it poses no risk and has just benefit.
But… nowadays few think in a systemic way, meaning considering the combination of factors instead of focusing on a single domain. AI is and will not be “living alone” disconnected from our world. Just think about a self-driving car; there AI has eyes, ears, wheels and a powerful engine under its total control.
In addition as AI is evolving at tremendous speed while other enabling technologies are as well (both at technical development and adoption level).
Thinking about the future of AI in a systemic way, meaning combining AI with technologies that are and will make use of it… it tells a more interesting story.
What does AI represent? Outsourcing our brain!
AI is mainly two things:
- software writing by itself (unders supervised or unsupervised conditions);
- software able to perform tasks that so far were “purely human” abilities.
AI may indeed represent the biggest outsourcing ever, the highest engineering creation of humanity and the most powerful tool ever built.
Yes, outsourcing, but not by a person to another, by a company to another, by a country to another as we are used to, but outsourcing from humans to an independent “creature” if it.
That’s the final motivation of AI, outsourcing the activities and related skills that distinguished so far humans from animals: brain rational activities.
AI by itself is like a brain outside of a body. You need to feed it with info (sensorial information) and then get “what it has to say about it”, that’s quite interesting already, as the list of tasks it can execute is amazing.
A nice list of events showing AI evolution can also be found here.
Let’s not indulge more on AI as plenty of valuable material is available, while in this article I’d like to focus on a systemic view, in other words: what becomes AI when combined with other technologies.
AI + IoT?
IoT stands for Internet of Things, what is it?
IoT can be seen essentially as the “eyes-ears” plus “arms-hands” of internet. Indeed it aims to create a wide, interconnected, interoperable network of sensors (detecting everything from sound, movements, presence detectors, vibrations, gases, liquids,…) and actuators (any electromechanical movement, activation of switches, …).
Let’s reconsider closer the self-driving car example, that’s AI + Sensors&Actuators. The sensors of the car include the cameras used by AI to “watch”, along with many other sensors feeding the car AI with real-time information about the engine, weels, transmissions,… Now the car AI not only “sees” and “think” but also it can “act” using indeed actuators, or in other words, AI is able to translate the decisions in actions accelerating, decelerating, turning, breaking, activating lights,… That’s a robot on 4 wheels.
But IoT is much more than that as it’s not restricted to a single car, it’s spread potentially all over the planet on an interconnected giant robot. In principle a sensor network spread throughout Europe coastal area providing information about wind can feed an AI to predict sailing conditions feeding in real-time the navigation bulletin; the output in this case is just informative, no action taken. But action can be in the loop as for instance in managing a network of power stations analysing real-time consumption on the grid while controlling as retrofit the power stations accordingly.
So now you can visualize it as a brain fed from multiple “senses” providing info to process while having multiple “hands” to act and both spread geographically, therefore being able to control multiple things.
The combination of these “senses” (sensors) and “hands” (actuators) combined with software to increase process efficiency is nothing really new but recently…:
- Senses (sensors) and Hands (Actuators) are getting more and more ubiquitous and interconnected, to the point that concerns are raising about security to prevent potentially highly damaging hackers attacks (Denial of Services and more);
- AI rewrite itself: that’s not a trivial difference to the existing programs. Most of the software running on automations (especially industry) privilege “proven” to “best” where not only the “proven” ones have been used for a long time and are predictable, but they also have at disposal plenty of software engineers knowing line by line the code and being able to identify bugs and fix them in short time. While AI relies on Neural Network created by itself. Although AI is trained (mostly), even the trainers don’t have idea about the internal structure of it. It’s a little disturbing, isn’t it?
In other words imaging extended networks of “eyes” and “hands” interconnected and spread worldwide connected to software that is writing itself. It’s certainly full of potential, but also raising some concerns that need to be addressed, as some people although trained and educated in school become hacker and therefore a threat, couldn’t be the same for trained AI?
AI + IoT + Internet
So far we have seen the combination of AI and IoT meaning giving AI execution power with no geographical limits. It’s like a “super-robot”.
Not let’s add “Internet” to the equation.
Internet is connecting computers and supercomputers all together. Clouds are also popping up like mushrooms and containing Virtual Servers. What are them? The software doesn’t need to stand and run “physically” on a single machine, but can run on multiple computers at the same time as if they were combined into a single computer (virtual servers indeed). It’s like this technology allowing a software to “fly” independently from the hardware.
That’s not the all story, Internet allows easily also “crowd computing”
Crowd Computing is a form of distributed work where tasks are handled by large numbers computers distributed across the internet. It means that a software can connect to several independent computers connected to internet, split complex calculations into token of work and assign them to the connected computers.
In other words, Crowd Computing allows a “brain” to split a “thinking process” into “parts” and assigning this “thinking parts” to many other computers while managing them all simultaneously, sort of. Therefore the computational power really skyrocket. A small electronic brain connected to internet in a “crowd computing” approach can easily become a huge supercomputer.
Often AI biggest limitation is considered computational power, but considering the above we already have a way to provide AI overcoming this computational power limit using a “systemic approach”.
So far we have a technology (AI) that can:
- execute tasks so far limited only to humans;
- write (and re-write) itself;
- connect to unlimited number of “super-senses” (as sensors are becoming more and more powerful at lesser and lesser cost) at light-speed and potentially spread all over the planet;
- connect to unlimited number of “super-hands” (as also actuators are becoming more and more powerful and widespread) at light-speed and potentially spread all over the planet;
- potentially extend its brain to any connected computer (the technology is there, if it will be used or not it’s up to “us”)
In addition, we should take into consideration that hackers have developed a talent to “possess” other computers over the network, hosting in them remote “agents” able to perform tasks. But… AI is proving good (and even superior to us humans) in so many typically human tasks, why shouldn’t AI be good (or even better) than hackers too in possessing computers extending its own “brain power”?
Mmmh, this should give us some thinking. Just thinking for now and maybe starting to define preemptive regulations and security procedures.
Finally, the current trend suggests that significant additional efforts will be devolved to make AI more and more powerful and independent enabling us to outsource more and more “brain work” while having AI increasingly interconnected with our reality watching it and acting on it.
That really makes me think about the possible consequences and probably really motivate us to start research groups on possible unwanted consequences and how to continue technology evolution while avoiding them.
Marco is a Digital Agriculture international expert with 20Y+ experience in leveraging digital innovations to the benefit of the market. He owns a Math degree & MBA, and before focusing on digital agriculture, he successfully worked with his teams to develop and bring to market several new technologies and products in the fields of environmental monitoring (low-cost air quality city monitoring), risk mitigation (Unesco Petra site) focusing on Internet of Things (IoT), advanced sensors, Big Data, predictive analytics, Artificial Intelligence.
He co-founded 3 companies receiving international recognition by the European Enterprise Network (innovation success in 2007), was mentioned in Forbes in 2008, was awarded the Stanford University “Best Startup Award” at the Italian Innovation Day in 2011, and was IBM Smarter Planet finalist and Global Entrepreneur.
Recent tangible achievements: saving of water up to 50% with increased production of 250% in semi-arid climate; 30% average pesticide reduction in orchards along with agronomic risk reduction.
Main customers/partners supported in Digital Agriculture: Nestlé, Syngenta, Netafim, Omya, Agroscope, Purdue University.