Future Applications: AI in Entertainment
In the previous two blogs of this mini-series, we have discussed current applications of artificial intelligence in entertainment. Today, we will cover possible future applications of artificial intelligence.
Using computer vision, computers can now process images and “see” things (using convolutional neural networks). Essentially, they are able to analyze images by putting numerical values on each pixel. Using this technology, people — or should I say, the computers — have been creating deepfake videos: videos where it takes one person’s appearance and puts it in place of another in videos (or images). Some people, however, are currently using this technology irresponsibly; they post inappropriate content online using the likeness of celebrities.
If this technology is developed even more and used responsibly, deepfake could be useful in fields like photography and moviemaking. Perhaps in the future, services will make you look like you were on the moon or at a significant historical event. It could also be used to adjust the quality of the image. It couldbasically become an expert in photoshop and video editing, if it were developed correctly; it would save a lot of time. Computers would be able to do this “editing” relatively quickly. There would be no need to spend an hour editing a single image.
A second possible development in the field of entertainment due to AI would be in chatbots. Now, chatbots do currently exist; however, there is definitely room for improvement. As Microsoft’s Tay taught us on Twitter, there needs to be restrictions on what the artificial intelligence in the chatbot should be able to learn. Some people may think a possible solution to that issue would be to add a filter; that is, give the chatbot a list of words that it cannot use or learn. The issues with that solution are that it is too tedious, there may be words that should be restricted but are not restricted, and sometimes the connotation of the word depends on the context.
As we have seen earlier before in this mini-series, the solution to an open-ended question with many possibilities typically involves AI. Instead of giving the system specific words to restrict, a potential solution for chatbots would be to train a neural network to identify complete sentences or phrases as positive or negative. Using that information, the chatbot will then decide whether or not to learn from that post: for example, if the model classifies the post as negative, you generally would not want to learn from that post.
The last potential use of artificial intelligence in entertainment I will cover in this blog is singing. Currently, devices like Siri and Alexa use text to speech and synthesize words. Using a combination of pre-recorded audio (concatenation synthesis) and computer-generated audio (formant synthesis), they are able to speak in a way that is understandable and not incredibly monotonously. In the future, if artificial intelligence for audio processing continues to improve, then they could even sing songs very well.
Concluding This Mini-Series
Overall, artificial intelligence is extremely applicable to all of us. Whether we are adults who enjoy online shopping, teens who play video games, or children who love to watch YouTube, we are all using artificial intelligence pretty much every day of our lives. In the future, this technology could be developed upon, and its uses can become even more noticeable, especially in the field of entertainment.
Stewart Kwok is a Student Ambassador in the Inspirit AI Student Ambassadors Program. Inspirit AI is a pre-collegiate enrichment program that exposes curious high school students globally to AI through live online classes. Learn more at https://www.inspiritai.com/.
