AI-Generated Content & the Human Touch: Preparing Your Teams for Emerging Tech
It feels like just yesterday that robots were a thing of sci-fi films and cartoons. But just like that, they’re now helping us write articles, create images, and produce videos. Granted, we’re not quite ready to hire them full-time, but we are getting a really good preview into some of the things we can expect Artificial Intelligence (AI) to bring to the digital production world. So we’re taking a pulse check on what we currently know about AI-generated content and how to prepare your teams for what’s to come.
The Most Popular AI on the Market for Marketers
When it comes to generating content and streamlining processes with AI, there are a few standout tools that designers and developers are exploring.
ChatGPT. If you haven’t heard of this tool, you’ve likely been living under a rock — considering we’re now even seeing counter apps that prevent students from using it to do their homework. The tool created by OpenAI uses the company’s GPT-3.5 (Generative Pre-Trained Transformer) model to generate people-like responses to text prompts, meaning it has to learn the structure and patterns of natural language before being fine-tuned for specific tasks. This training is provided through extensive data, from books and articles to online conversations, and is robust enough to write poems, articles, and essays, send messages on apps and social platforms, generate headlines, and more.
For Images & Video:
Midjourney. This generative AI program was created and hosted by a California-based independent research lab and runs entirely on a Discord server, making it available to millions of people without the need of specialized computer hardware. Unlike Dall-E 2 and other competitors in image generation, Midjourney offers a more artsy, dream-like style to requests. The tool asks you to input a worded prompt for an image and within seconds, you’ll have multiple attempts at the visual, which you can continue to refine through feedback.
Copilot. This programmer assistant by GitHub uses OpenAI Codex to suggest code and entire functions. Trained on billions of lines of code, the tool allows developers to turn natural language (aka English) into coding suggestions in real time, right from your editor.
CodeWhisperer. Very similar to Copilot, Amazon’s machine learning-powered generator is designed to assist developer productivity by producing code recommendations based on their comments written in natural language. It can provide multiple code suggestions in real-time, so as you write code, it provides suggestions based on that existing code and any comments you include.
Runway Gen2 (images)
Eleven Labs (voice overs)
SOUNDRAW AI Music (music)
The Importance of the Human Touch — Especially Now
While these tools are getting lots of attention, it’s important to note that everyone is still in the discovery phase. Sure, they’re available and ready to be used, but you shouldn’t expect any of them to lead a project or provide a deliverable all on its own. Here’s why.
As the tech continues to learn from technical data, it is still far from grasping the humanality and emotion that connects with readers and viewers on a personal level. For example, ChatGPT is known for often generating “plausible sounding but incorrect or nonsensical answers” with redundancies, broken English, and unpleasant verbosity.
Similar limitations apply for Midjourney. While it’s capable of producing unique images, its output relies heavily on human feedback and involvement to get to an iteration that meets today’s innovation standards. And even then, anyone with a creative eye can detect the robotics behind the visual.
If you’re going to start exploring these tools, it’s so important to note the valuable piece that the human touch still provides. As they are right now, AI-powered tools are helping us replace very technical tasks/skills, but they don’t offer the consciousness to analyze a problem and offer a strategic, creative solution to complex problems like its human counterparts. Creativity, imagination, and flexibility are things AI just can’t fulfill – at least for now!
Within the next couple of years, we can expect to see AI taking the load off developers in terms of needing specialized grammar knowledge for specific languages, but for now, that expertise matters for compelling and differentiable results.
Investing in the Future of AI
So with the benefits of AI-generated content still in somewhat of a gray area, what’s the best way to utilize these tools right now? Well, a lot of that depends on your distinct processes and people. Most prominently, we’re seeing tools like ChatGPT and MidJourney accelerate early phases of the design process, eliminating the mental hurdle a blank canvas can create.
“If we are creating a banner campaign from scratch, we’re tasked with gathering research, brainstorming ideas, imagining and drawing mockups, etc.,” Fabian Miranda, Studio Manager | Web Dev & Production at Assemble, explained. “But these tools can simply scour thousands of resources for references and proofs-of-concept related to any task for us. Essentially, they can serve as resource tools, which can expedite the design process significantly.”
And that’s not the only process that can benefit. These tools also offer a project management element that can streamline internal planning. “We’re seeing the value in using tools like ChatGPT to help us develop the scope of a project,” Fabian added. “Let’s say I need to create a headless website with 10 pages and a contact form. At first, we tested its ability to define a good modern stack given this information, but its ability to make these types of critical decisions just isn’t there — the options were vague and inconclusive. So instead, we kept that layer of the process to our team, and told ChatGPT the stack and backend that we planned to use. It was able to generate a surprisingly accurate timeline based on an estimation of hours and phases for each step of the process. Of course, it required tweaking and adjustments to fit our specific processes and client schedules, but we continued to test it on other projects, and overall, it has great potential to accurately assist with planning.”
What Your Teams Can Do Now
With everyone eager to catch the wave, the biggest thing your teams can do right now is start getting familiar with these technologies. Encourage each department to explore the ways in which they can integrate into your processes and gain experience on how to achieve the best outputs.
“At Assemble, we have our designers diving into MidJourney to see how they can get the best mockups for a concept they are working on,” Fabian shared. “When they need inspiration, they’re asking how they can get the best results and how descriptive they need to be? Can we use the tool to apply different types of illustration? Can we get something realistic, something that resembles a 3D rendering, or something that looks like a cartoon? The only way to get those answers now is to use the tools first-hand, experiment with their capabilities, try specific combinations of keywords in different prompts, and then refine/depurate, and so on.”
If you’re reluctant to throw your teams right into it, consider external training. Many higher education institutions and online platforms offer courses that your team members can take in-person or remotely to further educate themselves on the capabilities and opportunities afforded by these tools. Assemble, for example, has our team currently enrolled in a 7-week Artificial Intelligence with Python course through Harvard University.
“We’re exploring the concepts and algorithms at the foundation of modern artificial intelligence hands on, while diving into the ideas that give rise to technologies like game-playing engines, handwriting recognition, and machine translation,” Fabian explained. “We’re going to continue to expand our knowledge of these tools and explore all the ways it can elevate our creative processes to better serve our clients.”
At Assemble, we love staying on top of all the latest tech and tools in the digital production world. So to learn more about the capabilities of AI-generated content right now, and how your teams can both start taking advantage and preparing for what’s to come, contact our experts at Assemble today.