By Mervin See, Channel NewsAsia
Published: 19 Mar 2019 - 11:24 PM
In the Commission On Science And Technology For AI and Robotics (COSTAR), they discussed the rights and ethics of AI.
To Give Or Not To Give, The Rights and Ethics Of AI
India started off by suggesting that only Humanoid Robots should be granted rights. Saudi Arabia countered this by saying that there is a risk of not allowing all AIs to be granted rights, as they might protest back against them. Iran countered this, saying that as robots follow a strict code and cannot feel physical pain, this is wrong as they can still experience deep learning and feel emotional pain.
Next, COSTAR discussed AI rights and what constitutes them. Firstly, China defined Artificial Narrow Intelligence (ANI) as AI that are non-sentient and can perform simple tasks. In regards to laws, Saudi Arabia, Egypt and other countries agreed that as the creators, they should be charged for any crimes the AI commit as they created the code. ANI should not be granted rights as they are unable to absorb information. Secondly, Singapore defined Artificial Super Intelligence (ASI) as AI that surpasses humans in all capabilities. America called for a complete ban, as ASI are dangerous and the situation might get out of hand, they are weapons of mass destruction and can even lead to a violent rebellion against humans. However, the consensus supported by Kenya is to allow but limit the number of ASI as they can slip past our current restrictions and to set up fail-safes in case they go rogue.
Singapore then defined Artificial General Intelligence(AGI) as AI equal or exceeding human-level intelligence.
In terms of Ethics, the conclusion which was supported by the Democratic Republic Of Congo (DRC) was that if AGI commits a crime, both the Developer and the Robot are punished. Countries then talked about the extent of rights AGI should get. Turkey also said that if any discrimination or prejudice is shown by robots, they would be shut down immediately as it would create lines of division between races. The United Kingdom (UK) also stressed the importance of drawing a line between Humans and AI. However, countries like North Korea felt that they should not be granted rights as it would decrease efficiency and they are invented for the sole purpose of benefiting mankind, and we should use and maximise the benefits to the fullest extent.
The general consensus supported by the UK and China is that they should get limited rights.
Preventing Conflict- Appeasement or Restriction
Lithuania started off by suggesting use of propaganda and DRC suggesting giving AGI labour rights, restricting the amount of hours they can work to let them rest despite them not getting tired. However, America suggested creating labour unions to ensure that they keep civil rights.
These methods of appeasement will ensure happiness among AI and be a moral way of dealing with tensions.
However, Egypt and Kenya felt that as the creators, we have to restrict the rights of AI as they might rebel, they feel that regulation of ASI is important, as they can slip past our current restrictions. India and France also suggested restricting ASI’s intelligence and freedom of choice as it would otherwise be too unpredictable.This would prevent a civil war between man and machine. Furthermore, North Korea proposed restricting robots from creating other robots, as the abundance might be too hard to handle.
Thus, a two-pronged method between appeasement and restrictment, while giving them limited rights, might prevent conflict between man and machine.
Different Fail-Safe Methods:Prevention Vs Elimination
The method suggested by UK, Lithuania and North Korea was for humans to conduct regular check-ups, humans as a backup, rigorous testing of anomalies and stable testing.This will prevent malfunctions from happening as the robots might have prejudice or violent tendencies.
Mexico suggested a reward and punishment-based system where robots would be given rewards or punishments such as fines or jail terms based on their behavior. India felt that we can restrict communication or delete their memories when they are having rebellious thoughts while North Korea suggested an identification to allow tracking and a physical prevention of hacking. Prevention will stop robots from being able to rebel entirely, and will defeat the problem at its roots.
However, elimination of robots can happen with a physical kill switch or a system override. This would make an easy weakness that can be used to eliminate them. However, Saudi Arabia opposed as it would be immoral to use a kill switch as it is likened to killing a human being. These elimination methods can be used as a means to suppress and restrict any rebellion but may be too cruel.
In the end, Iran and Ukraine suggested a compromise to use both methods as fail-safe methods.
Thus, only AGI have similar rights and ethics to humans.
Are Robots Human Beings?