Artificial Intelligence is a very big and useful technology that has taken the world by storm. As beneficial as it is, it also has its drawbacks. No matter how efficient they are, machines only work on algorithms and these algorithms should not be over trusted for several reasons.
There may be a genuine error in its algorithm that leads to different actions from the expected or desired ones. A bug or malware may find its way into the code and this will lead to abnormal behavior of these machines. Programmers may deliberately input wrong codes for ulterior motives. This is why explainable artificial intelligence is the way to go.
With explainable artificial intelligence, every user will be able to understand how a machine works. Besides, the machines will come with a high level of transparency and accountability. Every machine should be able to explain why certain actions need to be taken to its users.
It should also explain why that is the best option and why other alternatives may not work out for a particular situation. Explainable artificial intelligence also aims at making it obvious to users when a particular machine has failed on a particular task and when it has succeeded in the task. It will make the users be in total control and not the machines.
Users should also be able to know when to trust the actions of these machines and when to verify further. And if an error is noticed, users should be able to fix the error without the intervention of any developer.
The goal of this human understanding of artificial intelligence is to put humans in total control so that no action is taken without human endorsement. So before any machine takes any action even if it feels it is the best action to take, the machine still has to seek user’s permission.
All these are necessary because of the dangers inherent in uncontrolled artificial intelligence. Sometimes a machine might plan to carry out a task but adopt a destructive method towards carrying out the task. As an illustration, you can tell a smart car to take you to a certain location as quickly as possible. In the bid to carry out your instruction to the letter, the car may disobey traffic lights and break some other traffic rules thereby causing serious accidents in the process.
This is not likely with explainable artificial intelligence because the car will explain its intention and seek your permission before the task is carried out. Of course, you won’t endorse such. You will rather ask for other better and safer options.
The biggest concern lies in military activities. There is no doubt, artificial intelligence is already being adopted in the military. In the nearest future, there will be robot soldiers that will be deployed during military actions. What if terrorists find a way to manipulate the algorithms of these robots? This could be very dangerous.
Instead of targeting the enemies, the robots could turn back and wipe out their very own team and also unleash destructive ammunitions against the civilians of their own country. Apart from that, there could be some disgruntled soldiers who are secret saboteurs. Have you ever wondered what will happen if a compromised robot soldier gets hold of a rocket launcher? The enormity of the destruction is better imagined.
The smart car scenario could also play out in military activities if there is no human control. If a couple of soldiers have been captured by the enemies and they are being conveyed away in a helicopter and a robot soldier spots the helicopter, what do you think will happen? It will blow up the helicopter without any hesitation because it has been programmed to attack enemies on sight.
With human control being in place, all these cannot happen. Before any machine takes any action, it will seek human endorsement except if the user is also a terrorist or a disgruntled saboteur. This is one of the reasons big names in science and technology like Bill Gates, Steve Wozniak, Elon Musk and Stephen Hawking have warned that artificial intelligence could spell doom to human race if regulatory and oversight function are not placed on the technology.
Without the intelligible and control aspect of explainable artificial intelligence, robots could get super intelligent and begin to change its codes. They can code themselves to stop listening to their users, they can code themselves to wipe out the human race so that only robots will exist. This is why there is a lot of concerns about the future of artificial intelligence. The best way to work against the concerns is to make algorithms simpler to understand, more transparent and more trustworthy to users and actors.
Another major concern is that all eventualities cannot be envisaged. Robots are programmed to take a certain decision depending on the situation on the ground. What will now happen if a situation that is not captured by the algorithms occurs? This is another kind of danger. As the saying goes, power is dangerous without control. Artificial intelligence is a real danger without proper control. This does not dispute the fact that artificial intelligence comes with a lot of benefits. In fact, it is about the most beneficial technology. Imagine how much companies save by purchasing a robot to do the jobs of about ten employees without going on leave, without earning monthly pay and without calling in sick for years. However, no matter how beneficial it is, if the cognitive necessity of explainable artificial intelligence is not taken seriously, this technology could mark the end of human race.
Due to these concerns, European Union gave the following recommendations for artificial intelligence.
Their most important recommendation is the creation of an agency for artificial intelligence and robotics. The agency will regulate both industries within European Union countries. EU parliament also recommends that there should be a clear and simple legal definition of smart autonomous robots . And their registration should be put in place to enable the registration of every robot. There should be a company responsible for the negative actions of any robot.
There should also be a comprehensive code of conduct for robotics engineers that guide the ethical aspect of the production, design and the use of robots. All production companies for AI and robotics need to send reports on how these two will contribute to the country’s economy. This is necessary for the purpose of social security and taxation. There should be a law that will mandate companies manufacturing robots and companies using them to purchase insurance policies to cover the damages caused by their robots.
Due to the concerns about looming mass unemployment, EU parliament recommends a general basic income for everyone as one of the ways to curb the mass unemployment likely to be caused by the implementation of explainable artificial intelligence.
