Risks with AI

All technology malfunctions, and when a technology is not very sophisticated or doesn't do anything of vital importance, this usually isn't too much of a problem. However, AI is a relatively advanced technology - even simple reactive AI - and will often be responsible for some important functions. 

For example, in 2016, Microsoft introduced a chatbot named Tay to Twitter, and by learning from the tweets she was interacting with, she quickly began spewing racist comments. She used the data from other users' comments to create tweets that could plausibly be from another user. Learn more about that incident in this link to Microsoft's blog.  

Many people may not trust AI because they don't understand it. In some cases, the mistrust may be justified. As mentioned earlier in this module, it's important that we put measures in place to ensure that AIs behave in a way that aligns with our values so that they don't start making decisions that are damaging to us.

One may consider, for instance, the addictive nature of certain social media platforms to be a consequence of damaging decisions made by the AI's algorithms, going a bit further than their creators originally intended.

AI is really good at automating certain processes, especially repetitive ones. This does mean, however, that implementing AI in a workplace may put some jobs at risk. There's a good chance that AI will make many jobs obsolete, but research has shown that AI does not work at its best if left unattended. This means that new job opportunities in the area of regulating and monitoring AI will emerge as AI becomes more prominent in the workplace.