It is popularly thought that so much work is being done in the field of machine learning and content creation, it is only a matter of time before a cinema release or TV programme is written and directed by an AI.
Here, Forbes has listed some already significant achievements in similar fields that would seem to pave the way to this eventual breakthrough
An AI computer was given hundreds of samples of 17th-century “Old Master” style paintings and was asked to create its own paintings. One of the paintings it came up with is titled “Portrait of Edmond De Belamy” and fetched $432,500 at a Christie’s auction.
Another AI computer was fed hundreds of modern pop songs and came up with its own song named “Blue Jeans and Bloody Tears” complete with lyrics.
After being shown videos of people dancing, one AI computer came up with its own dance moves for a robot.
An AI computer was fed a number of rap songs and proceeded to create its own original rap song.
A computer named “Shelley” (in honor of the author of Frankenstein) was fed 140,000 horror stories and then proceeded to create even more chilling ones.
An AI computer was given all the novels that Game of Thrones was based on and then proceeded to write five new chapters continuing the plot line. OpenAI GPT-2 (Generative Pre-trained Transformer 2) was given the task of writing alternative endings to Game of Thrones.
If AI can write entire stories, create visual images and music, those disciplines can be combined and the AI’s role extended, to emulating existing actors or creating new “stars” for a franchise.
The legal minefield of using AI to create content
This begs the question, who would be the “author” of a film created by an Artificial Intelligence? The Forbes journalist Schuyler Moore suggests that ..
The most likely result is that some person (including an entity) will be identified as the “author” of the work under the Copyright Act. Here are some possible choices, assuming each person is different:
- The person that wrote the software.
- The person that owns the computer.
- The person that uploaded the underlying material that the AI computer based its work on.
- The person that acquired the resulting work.
A similar issue will be who to sue by anyone aggrieved by the resulting work, such as for copyright infringement, right of publicity, slander, or defamation. Once again, it may not be clear which of the four alternatives listed above is the proper defendant.
Then there are the claims that will arise from disgruntled third parties.
For example, if an AI computer creates a new work by combining elements from uploaded information, then the uploaded information has been copied to some extent, giving rise to a copyright infringement claim. The classic distinction has been between copying the “expression” from another work (not allowed) versus copying the “idea” of another work (allowed), but can AI computers understand an “idea”?
A similar issue will be right of publicity lawsuits filed by celebrities that believe that the characters created by AI resemble them. For example, if AI creates a character that is a composite of Brad Pitt and Tom Cruise, can neither sue? Can both?