ChatGPT and Lensa are two popular tools that have everyone curious about whether they understand the AI behind them or not. AI is all around us, but more recently has become extremely popular with the release of these two big programs. While they appear to be innocuous and fun, these programs actually have many asking some big questions about the AI and how it’s being used.
Lensa: Going Below the Lens of Fun Pictures
Lensa is a program that uses AI to create a variety of images based on uploaded selfies. It turns one image into multiple avatars using AI image generation technology. Users can also remove imperfections and objects, create short movies, and do basic image adjustments. If you’ve been watching on social media, you’ll see people sharing their portfolio of AI-created avatars.
The Concern:
Sounds harmless and fun enough? Lensa actually generates the avatars using Stable Diffusion, an open-source AI model that can create images based on text prompts. Stable Diffusion is built up on LAION-5B, which includes a gigantic open-source data set built from scraping images off of the internet. Sounds simple, but when you consider the bias in so many photos between women vs. men on the internet, you can see how the AI model may skew images of females to be hyper-sexualized compared to men being more sophisticated and professional. Then, when you factor in all of the inappropriate images that exist online, you can start to see how a model like this is flawed and potentially dangerous. Their AI training data is filled with racist stereotypes, pornography, and explicit images of rape.
When it comes down to it, their data is biased, and that can have a big implication for the long run. All of the work that cultures have done to reduce bias, level the playing field, and create opportunity for all, regardless of gender, race, beliefs, etc., could be set back if AI models aren’t built thoughtfully.
OpenAI: Opening the Doors to Intelligent Text
OpenAI released the ChatGPT chatbot on November 30, a chatbot designed to mimic human conversation based on prompts. The possibilities of this tool are endless for creating written content from tweets to even this article. Millions have flooded to this program to test it out to see how well it can write a social media post, or a full-length essay. Many have tested it purely out of curiosity, but many see amazing potential in the tool in the long run.
The Concerns
Is OpenAI truly open source? Many argue the model isn’t open source and poses some real questions. AI still has a long way to go in removing bias in the written word and how it’s answering questions. Users of the tool need not check their brain at the door, but pay attention to what the model is spitting out. Before using any verbiage OpenAI and ChatGPT create, any user should intelligently comb it to make sure the information is accurate and reflects their values. The platform is still actively working to improve the model, but as with all AI, it’s a work in progress.
Even with these concerns, we’re seeing something new happening in the world around us. Regular people with no prior interest in AI, nor technical understanding of how these models are creating their results, are flocking to try them out. They’re not just modestly curious, but coming in droves to play with AI. We have to learn how to use these new tools well, but we’ll only continue to see more tools like Lensa and OpenAI become commonplace in our daily lives. AI isn’t going anywhere!