AI or Artificial Intelligence is a topic that inevitably shows up in modern conversation. Every day when you open up the newspaper or a tech magazine there will be something that pertains to AI conquering new boundaries. There is news of layoffs because some company has developed AI tools that can automate a lot of business processes and human effort is no longer as necessary. What is now in demand is AI operators who can appropriately feed a computer information on the requirements and get back responses. And there is no end in sight. AI market size is expected to reach $1,339 billion by 2030. On the flip side, about 400 million workers could be displaced worldwide during the same time period. Though there are these unfortunate job losses, there are also additional jobs in the market as AI continues to grow by leaps and bounds in other fields.
Adding a caveat here before I proceed. For the more technically savvy user, yes, I understand the difference between AI and ML (Machine Learning). Please bear with me when I say AI. I will keep on generalizing it for this article as it is the umbrella term for the rest of the lingos. Ok, back to the topic. Think about the medical field. AI is analyzing a lot of medical reports as well as X-rays and MRI images. With access to large sets of data, AI can scan through specific patient cases and use information from the database to triage medical test cases and identify the most probable cause of illness. Weather predictions are lot better because of AI models that can run through millions of parameters. Today we do not have to take an umbrella when the meteorologist says that it is going to be a sunny day outside (for those ‘just in case’ scenarios – how many times have we been told that it would be sunny, just for thunder to start rumbling and storm clouds to gather quickly before we can make our way back to the car?). That’s just the tip of the innovations that happened since the advent of AI.
ChatGPT created a kind of revolution for the common man. If it was human, and we were in 17th century Boston, it would surely be termed as witchcraft. Unless you just came out of a long hibernation, you have surely used or have at least heard of ChatGPT.
Today, I wanted to discuss controversies regarding human creativity and the supposed demise of it since the start of ChatGPT. Let’s take some creative fields and discuss what the impact of AI has in each of them.
Text Generation – the beginning of it all
When AI started generating text, scientists wanted to do more. In the beginning, it was mostly English sentences that were grammatically correct, but lacked the depth required for a good literary article. The knowledge base or corpus, as it is called in AI parlance, only contained a limited set of documents that was fed to it. Using this limited set, the machine would randomly place a potential next word in a sentence using either the word before it or a couple of words up until that point. Later, to give more context to the sentences, weights were assigned to each probable occurrence. It improved sentence generation a bit, but still was not perfect.
With the arrival of deep learning and the coming of large language models, AI started to have a huge corpus, and also by that time prompt processing (the conversion of natural language sentences to keywords) improved. ChatGPT is a product of this model. It is based on a conversational model where you can ask something in natural language and it will search its knowledge base for an appropriate answer.
Think about it for a while. If you can feed in all works of Shakespeare to GPT (acronym for generative pre-trained transformer), it can generate stories or articles that feel as though they were written by Shakespeare. This is because the only knowledge this model has is based on his works. Expand this model to have complete works for thousands of authors. Now, AI model can start generating complete stories based on acquired knowledge from a lot of these authors.
It’s a lot of power to general public as well as all literary community. Unfortunately, if used unchecked, instead of using AI resources as a starting model, users may resort to plagiarism and copy the entire AI writing as their own. This definitely has a big impact on creativity and what you will see is a repeat of old stories with a fresh coat of paint on.
AI is for Artists
AI is not limited to text generation. Initial attempts to generate images from text prompts were rudimentary and used a specialized algorithm to generate passable images. However, deep learning models have given a huge boost to this. Models like DALL-E, Midjourney, Stability AI or Google’s DeepMind can now generate images that are very close to reality. Not only that, the newest version of Google DeepMind Genie 3 can generate lifelike videos and let you be in the video and interact with it, more or less in real time. Which means that you can add events (like a sandstorm or an alien landing) in that video on the fly.
In an artist’s hand, this is a very powerful tool. They can let AI create whatever they imagine and get inspiration from it. Since the options are unlimited, it becomes an inspiration galore (or overload). What is good for an artist is also a curse when it comes to journalism. There have been incidents when images generated by AI have been presented as real-world photos to show what war did to a country.
Video generation is a different beast on its own. When you can gamify worlds, you can use dummy worlds to train soldiers, pilots or even firefighters and give them real-world practice without any real risk to human life. Any kind of world event can be simulated for the training.
AI in Music generation
AI is not limited to visual composition. It has also creeped its way into music generation. AI models from OpenAI like Musenet can generate harmonious music using ten different musical instruments simultaneously. It supports a multitude of genres and within each genre incorporates the styles of specific musicians. Not only can it generate music, it can add music to existing poetry as well. In the hands of an able musician, this can be very useful. They can try out different variations of music and decide which variation suits best before starting to compose. It can inspire the creation of a new kind of music.
Closing Thoughts
One of the major challenges of plagiarizing directly from AI is that you may be using copyrighted material without consent. Because AI relies on publicly available documents, which are not necessarily copyright-free, it gets trained on copyrighted content as well. A good example would be articles in newspapers. Since we do not know what source was used for generating the content, we may step into the legalities of copyright issues. This is a big problem, as it still needs to be decided who owns copyright for generated content. Is it the machine that owns content? Let’s hope the “machine” does not get too intelligent and start suing us.
So, where do we go from here? The trend we see now, AI is going to be increasing multifold. The technology for AI is advanced but still young. From definition of AI, it is still in its teens. At the final state, AI will be growing into self-aware machines. We are at the junction of an interesting stage. We have seen inception of AI and have seen the tremendous growth. Let’s see where it goes during our lifetime. One thing is for sure, the possibilities of AI is limitless. AI combined with human imagination is bound to open up new doors for creativity.