With Artificial Intelligence (AI) being more and more in the public eye, it’s time to set the facts straight and discuss some of the things which have been worrying (or even exciting) us in terms of the future of robotics and AI. Of course with any new development of technology there are going to be risks and benefits, and because of the nature of the research, a lot of the risks cannot be definitively determined and therefore are based on predictions of AI researchers and creators.
One of the dangers with AI is that the programming will need to be so infinitely exact. Two options arise when considering an AI’s potential programmed goal. Either the AI is programmed to carry out a horrific goal- an act with mass casualties for example, or the AI is programmed to carry out a positive or beneficial goal, but uses a process to achieve it which is destructive or damaging. If any goal set for the AI is not perfectly in line with the ideals that the programmer has, then they may end completing tasks that do more harm than good.
AI are set to become superintelligent beings. Some experts believe it could happen within decades whereas some are unconvinced that it will ever happen. However, after recent developments with the Sophia robot there are concerned about the AIs becoming overly intelligent. The dangerous part here is that intelligence is paired with a level on control. Every species is controlled by a more intelligent being, for example we humans have pets by controlling them and being more intelligent in order to train them etc. A crazy thought, but logically, AI superintelligence could result in our being controlled.
AI also give the power of being able to replicate identities to create believable audio and video. This can be dangerous in days where the media is so prevalent in everyone’s lives. They can match this with profile and credit card information that is sifted from the vast scrolls of data on the internet. This article shows the powers of imitating political figures and the damaging effects AI could have in this respect.
Relating back to our thoughts on AI programming, weaponizing Artificial intelligence can be dangerous due to misaligned goals or AI being held in the wrong hands. Artificial intelligence systems are goal focussed, and whereas you average machine cannot have a programmed ‘goal’, autonomous weapons can. This can be missiles which seek heat or other goal focussed weapons which can do a lot of damage, and more specific and localised damage than current weapons may be able to achieve.
Although many administrative and logging jobs may be taken over by artificial intelligence, many sectors will grow exponentially as a result of the technological advancements. Jobs in areas such as online and digital marketing, virtual reality entertainment and space-related careers will become far more important and prevalent.
Reearch into artificial intelligence has been wide and varied, with some researchers believing that AI could be the answer to a meat-free diet for all. This BBC article talks of how the Ai could reproduce products so similar to meats that we would not be able to tell whether there was animal produce involved in its production or not. Other articles speak of how AI could help us to treat cancer or infectious diseases, or even sight loss.
It is clear that Artificial Intelligence could bring both hope and devastation to the world, and we need to invest in research now in order to plan ahead and try to make the programming of AI being safe and predictable.