- Amazon is launching AI software that lets kids create stories with music and animation.
- AI-generated art has led to the creation of racist images in the past, so Amazon is taking extra precautions with its software.
- One AI expert said the new software could stifle creativity.
Amazon is launching a new artificial intelligence (AI) program that helps kids make art, and the company claims it won't create unsavory images.
Create with Alexa uses AI to let children build their own stories with graphics and background music, with finished animated stories displayed on Echo Show devices. However, mixing art and AI has led to controversy in the past, so Amazon is trying to ensure things go smoothly this time.
“We have put in place multiple guardrails to ensure the experience is safe,” Eshan Bhatnagar, head of product for Alexa AI, told Lifewire in an email interview. “From the get-go, we made sure to use carefully curated data sources that do not contain offensive or inappropriate content to train models. Second, we limit input prompts for story creation to pre-curated selections, ensuring we don’t prompt models with inappropriate inputs. Third, we integrate content filtering solutions to automatically identify and remove inappropriate content.”
Amazon suggests its new program can be used by kids to inspire creativity. When a child has told Alexa they're ready to create, they select a theme, which could be "space exploration," "underwater," or "enchanted forest." They then choose and name a character, choose a color theme, and pick adjectives like "silly," "happy," or "mysterious." Then Alexa creates a story, complete with sound and visual effects.
“It’s an experience where Alexa becomes your child’s co-creator,” Nico Bishop, a UX designer whose team builds the customer experience for Create with Alexa, wrote in a news release from Amazon. Children say phrases like, “Alexa, make a story,” follow prompts, and then Alexa creates a five-scene narrative based on the child’s preferences. The AI also composes an illustration, background music, and sound effects for each scene. The child’s choices of character, setting, and other variables dictate the story’s direction—and the company claims no two stories are the same.
Amazon said the software is made possible by innovations in conversational and generative AI that help build the narrative, scenery, and sound. A language model trained on human-written stories generates the text of the story.
A scene-generation model illustrates the story by selecting a background image created by AI or human artists. It also adds relevant objects from the narrative, such as a pirate, in appropriate locations, orients gestures like waving hands, and sets facial expressions.
Alexa’s AI sound and music generator produces relevant sounds and background music. The software draws from a library of instrumental parts, including artist-created chord progressions, harmonies, and rhythms. An AI musical-arrangement system is intended to ensure that all the pieces fit together.
Keeping It Safe
While Amazon clearly hopes that kids will create bedtime stories with its software, other types of AI-generated art have led to the creation of racist images in the past. Safety measures were a top priority when creating Amazon’s software, Bhatnagar noted, adding that Create with Alexa requires parental consent before it’s enabled on a device.
"We sought feedback from parents and their kids on a variety of topics when developing this product," he added. "Their inputs played an important role in shaping the experience."
AI generative art can help kids develop their creativity and imagination, Amey Dharwadker, a machine learning engineer at Facebook, told Lifewire via email. He said children could also learn about new styles and techniques through AI-generated art by seeing how artists from different cultures and eras might have approached a particular subject or style.
"This can give them new ideas for their own artwork and help them develop their own personal style," he added.
But Dharwadker said that working with AI generative art could result in kids needing to learn to be creative.
"With AI doing all the artistic work, children may not feel the need to practice their own drawing and painting skills," he added. "They might also not try to come up with new ideas since they could simply ask the AI programs what to do. Another disadvantage of kids using AI generative art is that they might not develop the skills to collaborate with others."