Beginning with a research project that began over five years ago, Jack decided to take a leap of faith by going to an event at the Toronto International Film Festival. Backed up by none other than himself and his emerging brand, Jack experienced the hard way that not all feedback is positive as he went back to the drawing board multiple times in order to sketch out a company with a unique idea.
Fast forward to 2016 and Jack Zhang is the owner of Greenlight Essentials; a company that uses unique artificial intelligence software in order to come up with plot frameworks for target markets within the entertainment sector. Forging a career off the back of a unique idea is never easy and Jack continues to try and push his software into new corners of the film industry but he is slowly gaining followers as he produces high quality movie trailers based entirely off AI frameworks.
This AI-focused development process led to Concourse Media and Productivity Media, Inc. (PMI) acquiring the rights to the screenplay for "Impossible Things".
Jack believes that all entertainment professionals should, at the very least, have an awareness of the importance of movie data and data analysis when it comes to creating screenplays and Greenlight Essentials promotes a better understanding of clearly outlined formulas when it comes to appealing to large markets.
In the following interview, Jack Zhang shares some of his experiences working as an entrepreneur, from his early ideas to where he sees himself in the future.
So tell us a bit about yourself. Where are you from originally?
I'm originally from China but I moved to Canada when I was 13. I ended up going to Vancouver high school and I later studied mathematics at the University of Waterloo in Ontario.
How did you move from mathematics to film?
I've always been fascinated by film and it has always been my passion even though I went to school for mathematics. I ended up combing both when I started my company. I founded Greenlight Essentials because I think there are a lot of things we can do in film by applying the kind of machinery and technology that wasn't available 5-10 years ago.
When did you start the company?
I started working on the project in 2011 and it took many years of research and developing but I finally went to a Toronto International Film Festival family event in 2014. I managed to reach out to industry professionals but the feedback wasn't great so I went back to the drawing board and redesigned the software to be more interactive and more interpretable. I then founded the corporation in 2015 although I've been working on the project as a whole for 5 years now.
Was your family supportive of you starting a company just like that?
Of course. Before I got some industry traction I was still at the stage of deciding whether or not to pursue a 'real career' or dedicate myself to starting up a company. My parents were always very supportive and after I got some attention from the industry and started getting some clients then became even more supportive. We had quite a few conversations about my future right after I decided to start the company and it just went on from there.
What are some of the challenges you faced starting up your business?
Everything is a challenge at first. Time costs, development money, pay salaries to your employees; it's a lot to handle. Thankfully the Canadian government is very generous and they help out start-up companies by taking some of the initial costs off our shoulders. Also, a lot of the film industry is against using technology. They say "we don't want a machine telling us how well the movie would do. We just rely on our gut feelings." While I do think we need gut feelings, it doesn't hurt to look at the problems from another perspective. You add more data points to your knowledge and you see the full picture.
Is the psychological barrier one of your biggest challenges?
Yes, that's one of the biggest challenges we're facing right now, but I think if we find the right way to avoid and solve problems using this technology then it becomes one of the best tools to help you write a better screenplay and help you better match audience demands in the marketplace.
CREATING A BRAND
Could you explain how the Greenlight Essentials works exactly?
Yes, and I'm super excited to speak about this. It's very practical but essentially we've managed to take theoretical ideas and apply them to real world problems. We provide a service for movie producers and investors who can make better investments by using big beta and artificial intelligence. We currently work with a few different companies. Usually someone comes to us with a project and a script. We then talk about suggested casts and directors and then we help them to identify which market is their target market; both domestically and internationally. They might only be targeting the UK or Mexico or going for the international market, depending on the subject matter.
Are there particular elements that vary for different markets? For example, do you utilize a different framework for the American market than you do for European, Indian, or Chinese films?
Yes, they vary across regions and depending on genre. For example, with horror films we could identify that in order to conquer the biggest possible market we need a ghost film that has a family relationship within it. That would improve our odds of success tremendously. We also have about twenty other plot genes, as we call them, which we have to include in the screenplay in order to make a killer story. We try to fit most of them into the trailer but we also work with other resources. There are definitely patterns and key plot genes for each region in the world.
Is there a list of them somewhere or is it a trade secret?
We have about 40,000 different genes in our database. I don't think anyone would like to go through all of them. It would take such a long time. They're not a secret though.
If a smaller filmmaker wants to use your service, is it affordable to them?
We have different models for different filmmakers. We were initially out to service the top notch market where filmmakers have deep pockets and we charged five digit fees on each project but we later realized that there's a huge demand in the smaller market right now. We're currently trying to make our application form available online and that would mean that people could do a lot of the process themselves in order to cut costs. We can make it available to everyone and it will be much more affordable.
UTILIZING THE TECHNOLOGY
How much of the workload is done by the AI system?
It doesn't write a script for you. It help you generate and develop ideas. It helps you come up with a building blocks of scripts but it cannot write a full script by itself. You are still the writer. AI is just gibberish. It's really good when it comes to problems and information but when it comes down to creative power, it's still very weak. We still need human creativity to hold these things together. But if you use AI in the right way, just like we did, you can get a lot of information that's not available, like five or ten years ago. We can use this AI and transform it into useful knowledge then give it writers or screenwriters. We are here to help them match their target market. That's what we're trying to do.
When you say that a script it co-written with the aid of AI, what percentage is AI and what percentage is human?
Well, before a single word is written, our AI is able to give us the basic framework. You might need to have a piano and a bathtub scene in a horror movie, for example, so before a single word is written the AI has come up with a plot point. This gives an idea of how the screenplay should look. This also means that AI doesn't honour the logic of a plot and how it flows throughout the movie. This is where our human creators have to go in and decide how we can connect all of these elements together. Human creatives take ideas from AI and then they create a readable script.
Do some writers feel like this takes away from the creative process a little?
We are only trying to give them a direction where to go. As creators they have total freedom to follow that direction or not. All we are doing is outlining where the landmines are in a minefield. We say "You maybe shouldn't have a horror film that's about vampires and do this, and you shouldn't have zombies and do that, but a ghost horror might work with this idea. This is how you could develop it and connect all of the other elements, according to the AI." As a creator, you have total freedom to do what you want and we encourage our creative team to come up with ideas that they can connect to the story.
Does the AI also tell you specific things, like the age of the cast and how they should look?
In the thousands of descriptions of movies available there are certain elements that do include age, race, and so on. The AI is able to access anything that relates to a potential theme within the particular film, although we are still trying to expand the database to include even more descriptions. There's an unlimited amount of descriptions that we can put in the database for the AI to work with and these could include hair colour, eye colour, and such things. This is currently being worked on though and these particular elements are no clearly available now, although they will be in the future.
At some point, would it be possible to incorporate facial recognition software into the database?
It's something we're looking into right now actually but it's going to take some time to bring an image dataset online. Right now we're mostly working with outstructured text data such as summaries of movies. Later we are going to incorporate the actual images from the movie. It's going to take a lot of data power and storage though and we currently don't have the ability to do it. We are definitely going to do it at some point though.
When it comes to diversity, how would the AI know to consider an Indian or Native American star when the past history of movies predominately features Caucasian leads?
While our database is quite massive, and we've collected films from as early as 1982, there are a lack of films out there with diverse casts. No horror films star an Indian lead actor, for example, and so it's really hard to the AI to tell if they will be successful based on that. If there's no data for the AI to look into they you can't really draw any insights from anything but the power of the AI is that we can look into very detailed information within films and use rigorous analysis to see the factors that truly draw in an audience. This is something humans struggle to do. We see the film as a whole but the AI can establish whether its success will be because of the actors, the plot, or something even more specific within the movie. AI and Big Beta help us navigate through the information and so they could tell us whether or not it would make a difference if the diversity of the cast was changed.
STIRRING UP INTEREST
Have you reached a point yet where you are able to make a film from scratch?
Not quite but everything is working really well so far. Anyone who has looked at our Facebook page recently knows that we don't have a film yet but we've been able to deliver teaser trailers that generate a lot of excitement. By using the algorithms that we have set in place we've been able to create teasers out of the content and elements that audiences want to see. The videos currently have thousands of views and comments. People are saying that they want to see the movie and they're tagging their friends. It's really exciting to see that.
That is very exciting. And does this happen organically?
Our page isn't totally organic as we do spend some money on ads but our delivery costs are so low that we can afford it. This means that we are able to work at one seventh of the industry average compared to everyone else working in distribution who is trying to advertise on Facebook.
How is it that you are able to keep advertising costs down?
We are able to keep the cost down because we engineer a film for a target audience that we already know before the film is shot. We can show them the content they want to see. So, for example, I can just advertise a segment to the target audience and if they like that part of the plot then they are likely to follow our page. If it's something they love then the supporters comes naturally. It's a confined approach and it can take a long time to build up especially as Facebook has reduced the organic reach of certain pages which is why we do spend some money on ads now. It's a way to get the word out there quicker. Without paying it's just going to be a bit slower for us to catch some momentum.
What has been your biggest surprise so far during the creation process?
Probably how much it cost to utilise the process and turn it into a film trailer. It's been a lot lower than I expected because we're using so many resources that are readily available to us. We do expect the cost to go up later down the line though as we want to create a movie that is commercial quality. When we finally make the movie, it will be much better quality than the teaser trailer we put out. The trailer is just an MVP.
What's an MVP?
A minimum value product. It's something we just put out there to see if people liked it or not, to get some market reaction. So before the film is even made we are able to test out the market and see who will like it and whether or not they like it or not. In the traditional creative process, you have to make the entire movie first before you see whether or not audiences like it. If they don't then you've wasted millions of dollars. So instead we spend around five hundred dollars to see if people like what the AI came up with. We push it out there on social media and sometimes we hit the jackpot. The risk is massively reduced and a lot of money is saved.
So what's the next step from here?
We are currently looking forward to working with a few producers and some big news is going to come out during September as we will have some big announcements on collaborating with industry-level producers and distributors. We are pushing to get to theatres but we will have to wait and see what happens. If our distributors end up thinking the same way we do then we can definitely push forwards but we're still not completely sure right now.