How ‘good’ is AI?
Artificial Intelligence have brought the world into a new level of comfort. Just open up your mobile phone, and call Siri or Bixbi -they are your one and only AI secretary. If you wish to view a video that would interest you without searching a specific keyword, the algorithm of YouTube will simply recommend a video that you would like. Simple and useful.
Then, can we simply assume that AI is good? Of course no. There are problems that may arose from Artificial Intelligence (AI) would frighten you when you know it.
Privacy and Fraud
One major problem that numerous experts state is related to privacy issue. For instance, a GAN model was utilized to create faces that does not exist. When you go into this website, it will generate numerous images of faces that does not exist, however might exist.
What if one attempts to live with a fake identity in the internet? That person can simply create a face utilizing GAN model and commit illegal acts such as fraud. Can human simply determine whether or not this person is real? I would say impossible.
Another issue with privacy is the fact that people are training AIs from major companies without knowing they are giving the data. When one login to a new computer, he or she will probably come across Google asking if they are a human and choose pictures that has certain objects. When the user inputs their answers, they are unknowingly providing information to AI on image detection and sensing.
Creativity?
Our group have also generated a GAN model to create manga covers that might exist from Shonen Jump — one of the biggest manga publishers in Japan.
Another problem besides privacy issue was noticed from this work: these covers are very similar to the style of Shonen Jump. In other words, human artists can copy ideas from AI to create works.
If artists start utilizing AI to assist their lack of creativity, in further years the need for human to utilize creativity will decrease as people become more reliant on AI. Even nowadays, we can find works in fields of art generated by AI.
Listen to this music:
This is a music composed by AI. This is an extraordinary field of AI and a great jump of technology. However, we have to also notice the risks that make come across. AI in the future might take over humans’ job even in the field that requires creativity.
Prejudice and Biases
Now, we are aware that AI can be artists to certain level. Then, the question arose, can AI also detect emotion and react to it? Currently, many people assume that AI cannot detect emotion like humans and no computer or programs will ever be able to do so.
However, the reality is different. “While humans might currently have the upper hand on reading emotions, machines are gaining ground using their own strengths” stated MIT’s editor, Meredith Somers.
To test if AI has to power to do so, our group have designed an AI to detect humans’ facial emotion. Even though our group had limited amount of data and emotion groups of only six, the AI can precisely guess the users’ emotion from their facial expression. Artificial Intelligence are becoming more humane.
Then, can we simply congratulate this development? Last problem I would like to point out is related to biases that may arose from AI. We have to note that even if AIs are very humane, it is still programmed and designed by programmers that are human. This can lead to AIs holding bias and ideas that the programmers have. For instance, the voice of Siri and Bixbi are designed to be females, creating gender and job bias that secretaries are females rather than males.
AI will always follow human to help our lives to become more convenient. However, we have to also note that AI will execute what the programmers intend the AI to do, which can threaten our daily lives. AI will be good as long as we use it good.
