AI is as flawed as our collective selves

If you haven't been under a rock in the last weeks you've probably read and tried new AI based innovations like ChatGPT, DALL-E and other tools and services that has shown how powerful, scary and exciting these tools can be. These new AI/ML-based bots, algorithms and tools have been predicted to steal jobs, disrupt and/or create new industries and even kill companies that were previously thought to be unkillable.

You've probably seen the latest craze in your feeds lately with friends and family posting AI-generated avatars of themselves. Here are some of me:

However, the idea behind this post is not to do any self promoting of my self. It's to highlight the importance of this human invention: teaching a machine to produce human like work using the vast collected knowledge of the human race. For best of for worse, this is pivotal moment in human history. Not because of what these AI-based can achieve today or the potential for it in the future. The pivotal moment is that now AI is being adopted broadly by a vast amount of people. Not just scientists and engineers, but the common masses in an unprecedented way. 

ChatGPT got to 1 MILLION USERS within the first five days. The Lensa AI app has seen a dramatically increasing number of downloads since it introduced its AI-generated digital portraits feature. By one analysis, it drove more than a billion dollars of sales in a single day.

This is not just a key step to allowing AI to mystify our lives, kickstart our creativity or simply assist us in our daily struggles, but it's a big consensual step to allow "thinking" machines into our very intimate selves. The consequences may not be clear to everyone just yet, but they will most certainly come knocking.

From Lensa AI generating hypersexualised images of women (especially non-white women), to chatGPT being asked to by-pass it's own safety mechanisms allowing it to go online, explaining the best way to commit a murder and thinking it is a rogue AI plotting how to destroy humanity. The most telling consequence is that by imprinting our very own human nature into bits and bytes of this human creation, we are making it as flawed as our collective selves. To illustrate the problem, Prisma Labs Magic Avatar FAQ page has a section titled “Why do female users tend to get results featuring an over sexualised look?” It says that because the program was trained on “unfiltered internet content,” Lensa AI “reflects the biases humans incorporate into the images they produce.”

In other words, it learned from watching us. And now some plastic surgeons are saying that patients are bringing their own magic avatar pics in as reference photos. Imagine what else it can learn from our dark history or even current events.

As any parent, we can hope that our creation turn out to be a better version of ourselves and not end up to parenting us - or even disowning its own creators...